From d253a3182859c7c989449122a60d5f362f19ded0 Mon Sep 17 00:00:00 2001
From: "Kartik K. Agaram" <vc@akkartik.com>
Date: Tue, 9 Nov 2021 08:12:11 -0800
Subject: rename grapheme to code-point-utf8

Longer name, but it doesn't lie. We have no data structure right now for
combining multiple code points. And it makes no sense for the notion of
a grapheme to conflate its Unicode encoding.
---
 linux/126write-int-decimal.subx  |   4 +-
 linux/304screen.subx             |  14 +-
 linux/305keyboard.subx           |   6 +-
 linux/400.mu                     |   8 +-
 linux/403unicode.mu              | 146 +++++++--------
 linux/405screen.mu               | 390 +++++++++++++++++++--------------------
 linux/407right-justify.mu        |   2 +-
 linux/411string.mu               |  20 +-
 linux/apps/arith.mu              |  46 ++---
 linux/apps/parse-int.mu          |   2 +-
 linux/apps/print-file.mu         |   4 +-
 linux/apps/tui.mu                |   2 +-
 linux/browse/main.mu             |  40 ++--
 linux/browse/paginated-screen.mu | 176 +++++++++---------
 linux/mu                         | Bin 614607 -> 614614 bytes
 linux/mu.subx                    |  10 +-
 linux/tile/box.mu                |   4 +-
 linux/tile/environment.mu        | 114 ++++++------
 linux/tile/gap-buffer.mu         | 188 +++++++++----------
 linux/tile/grapheme-stack.mu     | 124 ++++++-------
 linux/tile/main.mu               |   6 +-
 linux/tile/surface.mu            |  18 +-
 linux/tile/value.mu              |  12 +-
 linux/tile/word.mu               |  48 ++---
 linux/vocabulary.md              |  12 +-
 25 files changed, 698 insertions(+), 698 deletions(-)

(limited to 'linux')

diff --git a/linux/126write-int-decimal.subx b/linux/126write-int-decimal.subx
index 04f8b021..c48e17c8 100644
--- a/linux/126write-int-decimal.subx
+++ b/linux/126write-int-decimal.subx
@@ -303,7 +303,7 @@ test-write-int32-decimal-negative-multiple-digits:
     # . end
     c3/return
 
-decimal-digit?:  # c: grapheme -> result/eax: boolean
+decimal-digit?:  # c: code-point-utf8 -> result/eax: boolean
     # . prologue
     55/push-ebp
     89/copy                         3/mod/direct    5/rm32/ebp    .           .             .           4/r32/esp   .               .                 # copy esp to ebp
@@ -402,7 +402,7 @@ test-decimal-digit-above-9:
     81          0/subop/add         3/mod/direct    4/rm32/esp    .           .             .           .           .               0xc/imm32         # add to esp
     c3/return
 
-to-decimal-digit:  # in: grapheme -> out/eax: int
+to-decimal-digit:  # in: code-point-utf8 -> out/eax: int
     # . prologue
     55/push-ebp
     89/copy                         3/mod/direct    5/rm32/ebp    .           .             .           4/r32/esp   .               .                 # copy esp to ebp
diff --git a/linux/304screen.subx b/linux/304screen.subx
index 0b24fdbf..fb8ac0ea 100644
--- a/linux/304screen.subx
+++ b/linux/304screen.subx
@@ -157,8 +157,8 @@ $print-stream-to-real-screen:end:
     5d/pop-to-ebp
     c3/return
 
-# print a grapheme in utf-8 (only up to 4 bytes so far)
-print-grapheme-to-real-screen:  # c: grapheme
+# print a code-point-utf8 in utf-8 (only up to 4 bytes so far)
+print-code-point-utf8-to-real-screen:  # c: code-point-utf8
     # . prologue
     55/push-ebp
     89/<- %ebp 4/r32/esp
@@ -170,31 +170,31 @@ print-grapheme-to-real-screen:  # c: grapheme
     8a/byte-> *(ebp+8) 0/r32/al
     # if (curr == 0) return
     3d/compare-eax-and 0/imm32
-    74/jump-if-= $print-grapheme-to-real-screen:end/disp8
+    74/jump-if-= $print-code-point-utf8-to-real-screen:end/disp8
     #
     (print-byte-to-real-screen %eax)
     # curr = *(ebp+9)
     8a/byte-> *(ebp+9) 0/r32/al
     # if (curr == 0) return
     3d/compare-eax-and 0/imm32
-    74/jump-if-= $print-grapheme-to-real-screen:end/disp8
+    74/jump-if-= $print-code-point-utf8-to-real-screen:end/disp8
     #
     (print-byte-to-real-screen %eax)
     # curr = *(ebp+10)
     8a/byte-> *(ebp+0xa) 0/r32/al
     # if (curr == 0) return
     3d/compare-eax-and 0/imm32
-    74/jump-if-= $print-grapheme-to-real-screen:end/disp8
+    74/jump-if-= $print-code-point-utf8-to-real-screen:end/disp8
     #
     (print-byte-to-real-screen %eax)
     # curr = *(ebp+11)
     8a/byte-> *(ebp+0xb) 0/r32/al
     # if (curr == 0) return
     3d/compare-eax-and 0/imm32
-    74/jump-if-= $print-grapheme-to-real-screen:end/disp8
+    74/jump-if-= $print-code-point-utf8-to-real-screen:end/disp8
     #
     (print-byte-to-real-screen %eax)
-$print-grapheme-to-real-screen:end:
+$print-code-point-utf8-to-real-screen:end:
     # . restore registers
     58/pop-to-eax
     # . epilogue
diff --git a/linux/305keyboard.subx b/linux/305keyboard.subx
index 32159e49..21be1081 100644
--- a/linux/305keyboard.subx
+++ b/linux/305keyboard.subx
@@ -121,15 +121,15 @@ $enable-keyboard-type-mode:end:
 
 # read keys or escapes up to 4 bytes
 #
-# fun fact: terminal escapes and graphemes in utf-8 don't conflict!
-# - in graphemes all but the first/lowest byte will have a 1 in the MSB (be
+# fun fact: terminal escapes and code-point-utf8s in utf-8 don't conflict!
+# - in code-point-utf8s all but the first/lowest byte will have a 1 in the MSB (be
 #   greater than 0x7f)
 # - in escapes every byte will have a 0 in the MSB
 # the two categories overlap only when the first/lowest byte is 0x1b or 'esc'
 #
 # Only use this in immediate mode; in type (typewriter) mode 4 bytes may get
 # parts of multiple keys.
-read-key-from-real-keyboard:  # -> result/eax: grapheme
+read-key-from-real-keyboard:  # -> result/eax: code-point-utf8
     # . prologue
     55/push-ebp
     89/<- %ebp 4/r32/esp
diff --git a/linux/400.mu b/linux/400.mu
index c757c970..a391a76e 100644
--- a/linux/400.mu
+++ b/linux/400.mu
@@ -113,8 +113,8 @@ sig skip-until-close-paren line: (addr stream byte)
 #sig skip-until-close-paren-in-slice curr: (addr byte), end: (addr byte) -> _/eax: (addr byte)
 sig write-stream-data f: (addr buffered-file), s: (addr stream byte)
 sig write-int32-decimal out: (addr stream byte), n: int
-sig decimal-digit? c: grapheme -> _/eax: boolean
-sig to-decimal-digit in: grapheme -> _/eax: int
+sig decimal-digit? c: code-point-utf8 -> _/eax: boolean
+sig to-decimal-digit in: code-point-utf8 -> _/eax: int
 # bad name alert
 # next-word really tokenizes
 # next-raw-word really reads whitespace-separated words
@@ -159,7 +159,7 @@ sig move-cursor-on-real-screen row: int, column: int
 sig print-string-to-real-screen s: (addr array byte)
 sig print-slice-to-real-screen s: (addr slice)
 sig print-stream-to-real-screen s: (addr stream byte)
-sig print-grapheme-to-real-screen c: grapheme
+sig print-code-point-utf8-to-real-screen c: code-point-utf8
 sig print-int32-hex-to-real-screen n: int
 sig print-int32-hex-bits-to-real-screen n: int, bits: int
 sig print-int32-decimal-to-real-screen n: int
@@ -174,7 +174,7 @@ sig hide-cursor-on-real-screen
 sig show-cursor-on-real-screen
 sig enable-keyboard-immediate-mode
 sig enable-keyboard-type-mode
-sig read-key-from-real-keyboard -> _/eax: grapheme
+sig read-key-from-real-keyboard -> _/eax: code-point-utf8
 sig read-line-from-real-keyboard out: (addr stream byte)
 sig open filename: (addr array byte), write?: boolean, out: (addr handle buffered-file)
 sig populate-buffered-file-containing contents: (addr array byte), out: (addr handle buffered-file)
diff --git a/linux/403unicode.mu b/linux/403unicode.mu
index 8594615a..655cae2b 100644
--- a/linux/403unicode.mu
+++ b/linux/403unicode.mu
@@ -1,31 +1,31 @@
 # Helpers for Unicode.
 #
-# Mu has no characters, only code points and graphemes.
+# Mu has no characters, only code points and code-point-utf8s.
 # Code points are the indivisible atoms of text streams.
 #   https://en.wikipedia.org/wiki/Code_point
 # Graphemes are the smallest self-contained unit of text.
 # Graphemes may consist of multiple code points.
 #
-# Mu graphemes are always represented in utf-8, and they are required to fit
+# Mu code-point-utf8s are always represented in utf-8, and they are required to fit
 # in 4 bytes.
 #
-# Mu doesn't currently support combining code points, or graphemes made of
+# Mu doesn't currently support combining code points, or code-point-utf8s made of
 # multiple code points. One day we will.
 # On Linux, we also don't currently support code points that translate into
-# multiple or wide graphemes. (In particular, Tab will never be supported.)
+# multiple or wide code-point-utf8s. (In particular, Tab will never be supported.)
 
 # transliterated from tb_utf8_unicode_to_char in https://github.com/nsf/termbox
 # https://wiki.tcl-lang.org/page/UTF%2D8+bit+by+bit explains the algorithm
-fn to-grapheme in: code-point -> _/eax: grapheme {
+fn to-utf8 in: code-point -> _/eax: code-point-utf8 {
   var c/eax: int <- copy in
   var num-trailers/ecx: int <- copy 0
   var first/edx: int <- copy 0
-  $to-grapheme:compute-length: {
+  $to-utf8:compute-length: {
     # single byte: just return it
     compare c, 0x7f
     {
       break-if->
-      var g/eax: grapheme <- copy c
+      var g/eax: code-point-utf8 <- copy c
       return g
     }
     # 2 bytes
@@ -34,7 +34,7 @@ fn to-grapheme in: code-point -> _/eax: grapheme {
       break-if->
       num-trailers <- copy 1
       first <- copy 0xc0
-      break $to-grapheme:compute-length
+      break $to-utf8:compute-length
     }
     # 3 bytes
     compare c, 0xffff
@@ -42,7 +42,7 @@ fn to-grapheme in: code-point -> _/eax: grapheme {
       break-if->
       num-trailers <- copy 2
       first <- copy 0xe0
-      break $to-grapheme:compute-length
+      break $to-utf8:compute-length
     }
     # 4 bytes
     compare c, 0x1fffff
@@ -50,7 +50,7 @@ fn to-grapheme in: code-point -> _/eax: grapheme {
       break-if->
       num-trailers <- copy 3
       first <- copy 0xf0
-      break $to-grapheme:compute-length
+      break $to-utf8:compute-length
     }
     # more than 4 bytes: unsupported
     # TODO: print to stderr
@@ -65,7 +65,7 @@ fn to-grapheme in: code-point -> _/eax: grapheme {
     }
   }
   # emit trailer bytes, 6 bits from 'in', first two bits '10'
-  var result/edi: grapheme <- copy 0
+  var result/edi: code-point-utf8 <- copy 0
   {
     compare num-trailers, 0
     break-if-<=
@@ -87,16 +87,16 @@ fn to-grapheme in: code-point -> _/eax: grapheme {
   return result
 }
 
-# single-byte code point have identical graphemes
-fn test-to-grapheme-single-byte {
+# single-byte code point have identical code-point-utf8s
+fn test-to-utf8-single-byte {
   var in-int/ecx: int <- copy 0
   {
     compare in-int, 0x7f
     break-if->
     var in/eax: code-point <- copy in-int
-    var out/eax: grapheme <- to-grapheme in
+    var out/eax: code-point-utf8 <- to-utf8 in
     var out-int/eax: int <- copy out
-    check-ints-equal out-int, in-int, "F - test-to-grapheme-single-byte"
+    check-ints-equal out-int, in-int, "F - test-to-utf8-single-byte"
     in-int <- increment
     loop
   }
@@ -104,55 +104,55 @@ fn test-to-grapheme-single-byte {
 
                                                               # byte       | byte      | byte      | byte
 # smallest 2-byte utf-8
-fn test-to-grapheme-two-bytes-min {
+fn test-to-utf8-two-bytes-min {
   var in/eax: code-point <- copy 0x80                         #                                 10     00-0000
-  var out/eax: grapheme <- to-grapheme in
+  var out/eax: code-point-utf8 <- to-utf8 in
   var out-int/eax: int <- copy out
-  check-ints-equal out-int, 0x80c2, "F - to-grapheme/2a"      #                         110 0-0010  10 00-0000
+  check-ints-equal out-int, 0x80c2, "F - to-utf8/2a"      #                         110 0-0010  10 00-0000
 }
 
 # largest 2-byte utf-8
-fn test-to-grapheme-two-bytes-max {
+fn test-to-utf8-two-bytes-max {
   var in/eax: code-point <- copy 0x7ff                        #                             1-1111     11-1111
-  var out/eax: grapheme <- to-grapheme in
+  var out/eax: code-point-utf8 <- to-utf8 in
   var out-int/eax: int <- copy out
-  check-ints-equal out-int, 0xbfdf, "F - to-grapheme/2b"      #                         110 1-1111  10 11-1111
+  check-ints-equal out-int, 0xbfdf, "F - to-utf8/2b"      #                         110 1-1111  10 11-1111
 }
 
 # smallest 3-byte utf-8
-fn test-to-grapheme-three-bytes-min {
+fn test-to-utf8-three-bytes-min {
   var in/eax: code-point <- copy 0x800                        #                            10-0000     00-0000
-  var out/eax: grapheme <- to-grapheme in
+  var out/eax: code-point-utf8 <- to-utf8 in
   var out-int/eax: int <- copy out
-  check-ints-equal out-int, 0x80a0e0, "F - to-grapheme/3a"    #              1110 0000  10 10-0000  10 00-0000
+  check-ints-equal out-int, 0x80a0e0, "F - to-utf8/3a"    #              1110 0000  10 10-0000  10 00-0000
 }
 
 # largest 3-byte utf-8
-fn test-to-grapheme-three-bytes-max {
+fn test-to-utf8-three-bytes-max {
   var in/eax: code-point <- copy 0xffff                       #                   1111     11-1111     11-1111
-  var out/eax: grapheme <- to-grapheme in
+  var out/eax: code-point-utf8 <- to-utf8 in
   var out-int/eax: int <- copy out
-  check-ints-equal out-int, 0xbfbfef, "F - to-grapheme/3b"    #              1110 1111  10 11-1111  10 11-1111
+  check-ints-equal out-int, 0xbfbfef, "F - to-utf8/3b"    #              1110 1111  10 11-1111  10 11-1111
 }
 
 # smallest 4-byte utf-8
-fn test-to-grapheme-four-bytes-min {
+fn test-to-utf8-four-bytes-min {
   var in/eax: code-point <- copy 0x10000                      #                 1-0000     00-0000     00-0000
-  var out/eax: grapheme <- to-grapheme in
+  var out/eax: code-point-utf8 <- to-utf8 in
   var out-int/eax: int <- copy out
-  check-ints-equal out-int, 0x808090f0, "F - to-grapheme/4a"  # 1111-0 000  10 01-0000  10 00-0000  10 00-0000
+  check-ints-equal out-int, 0x808090f0, "F - to-utf8/4a"  # 1111-0 000  10 01-0000  10 00-0000  10 00-0000
 }
 
 # largest 4-byte utf-8
-fn test-to-grapheme-four-bytes-max {
+fn test-to-utf8-four-bytes-max {
   var in/eax: code-point <- copy 0x1fffff                     #        111     11-1111     11-1111     11-1111
-  var out/eax: grapheme <- to-grapheme in
+  var out/eax: code-point-utf8 <- to-utf8 in
   var out-int/eax: int <- copy out
-  check-ints-equal out-int, 0xbfbfbff7, "F - to-grapheme/4b"  # 1111-0 111  10 11-1111  10 11-1111  10 11-1111
+  check-ints-equal out-int, 0xbfbfbff7, "F - to-utf8/4b"  # 1111-0 111  10 11-1111  10 11-1111  10 11-1111
 }
 
-# read the next grapheme from a stream of bytes
-fn read-grapheme in: (addr stream byte) -> _/eax: grapheme {
+# read the next code-point-utf8 from a stream of bytes
+fn read-code-point-utf8 in: (addr stream byte) -> _/eax: code-point-utf8 {
   # if at eof, return EOF
   {
     var eof?/eax: boolean <- stream-empty? in
@@ -162,18 +162,18 @@ fn read-grapheme in: (addr stream byte) -> _/eax: grapheme {
   }
   var c/eax: byte <- read-byte in
   var num-trailers/ecx: int <- copy 0
-  $read-grapheme:compute-length: {
+  $read-code-point-utf8:compute-length: {
     # single byte: just return it
     compare c, 0xc0
     {
       break-if->=
-      var g/eax: grapheme <- copy c
+      var g/eax: code-point-utf8 <- copy c
       return g
     }
     compare c, 0xfe
     {
       break-if-<
-      var g/eax: grapheme <- copy c
+      var g/eax: code-point-utf8 <- copy c
       return g
     }
     # 2 bytes
@@ -181,23 +181,23 @@ fn read-grapheme in: (addr stream byte) -> _/eax: grapheme {
     {
       break-if->=
       num-trailers <- copy 1
-      break $read-grapheme:compute-length
+      break $read-code-point-utf8:compute-length
     }
     # 3 bytes
     compare c, 0xf0
     {
       break-if->=
       num-trailers <- copy 2
-      break $read-grapheme:compute-length
+      break $read-code-point-utf8:compute-length
     }
     # 4 bytes
     compare c, 0xf8
     {
       break-if->=
       num-trailers <- copy 3
-      break $read-grapheme:compute-length
+      break $read-code-point-utf8:compute-length
     }
-$read-grapheme:abort: {
+$read-code-point-utf8:abort: {
       # TODO: print to stderr
       print-string-to-real-screen "utf-8 encodings larger than 4 bytes are not yet supported. First byte seen: "
       var n/eax: int <- copy c
@@ -208,7 +208,7 @@ $read-grapheme:abort: {
     }
   }
   # prepend trailer bytes
-  var result/edi: grapheme <- copy c
+  var result/edi: code-point-utf8 <- copy c
   var num-byte-shifts/edx: int <- copy 1
   {
     compare num-trailers, 0
@@ -225,48 +225,48 @@ $read-grapheme:abort: {
   return result
 }
 
-fn test-read-grapheme {
+fn test-read-code-point-utf8 {
   var s: (stream byte 0x30)
   var s2/ecx: (addr stream byte) <- address s
   write s2, "aΒc世d界e"
-  var c/eax: grapheme <- read-grapheme s2
+  var c/eax: code-point-utf8 <- read-code-point-utf8 s2
   var n/eax: int <- copy c
-  check-ints-equal n, 0x61, "F - test grapheme/0"
-  var c/eax: grapheme <- read-grapheme s2
+  check-ints-equal n, 0x61, "F - test code-point-utf8/0"
+  var c/eax: code-point-utf8 <- read-code-point-utf8 s2
   var n/eax: int <- copy c
-  check-ints-equal n, 0x92ce/greek-capital-letter-beta, "F - test grapheme/1"
-  var c/eax: grapheme <- read-grapheme s2
+  check-ints-equal n, 0x92ce/greek-capital-letter-beta, "F - test code-point-utf8/1"
+  var c/eax: code-point-utf8 <- read-code-point-utf8 s2
   var n/eax: int <- copy c
-  check-ints-equal n, 0x63, "F - test grapheme/2"
-  var c/eax: grapheme <- read-grapheme s2
+  check-ints-equal n, 0x63, "F - test code-point-utf8/2"
+  var c/eax: code-point-utf8 <- read-code-point-utf8 s2
   var n/eax: int <- copy c
-  check-ints-equal n, 0x96b8e4, "F - test grapheme/3"
-  var c/eax: grapheme <- read-grapheme s2
+  check-ints-equal n, 0x96b8e4, "F - test code-point-utf8/3"
+  var c/eax: code-point-utf8 <- read-code-point-utf8 s2
   var n/eax: int <- copy c
-  check-ints-equal n, 0x64, "F - test grapheme/4"
-  var c/eax: grapheme <- read-grapheme s2
+  check-ints-equal n, 0x64, "F - test code-point-utf8/4"
+  var c/eax: code-point-utf8 <- read-code-point-utf8 s2
   var n/eax: int <- copy c
-  check-ints-equal n, 0x8c95e7, "F - test grapheme/5"
-  var c/eax: grapheme <- read-grapheme s2
+  check-ints-equal n, 0x8c95e7, "F - test code-point-utf8/5"
+  var c/eax: code-point-utf8 <- read-code-point-utf8 s2
   var n/eax: int <- copy c
-  check-ints-equal n, 0x65, "F - test grapheme/6"
+  check-ints-equal n, 0x65, "F - test code-point-utf8/6"
 }
 
-fn read-grapheme-buffered in: (addr buffered-file) -> _/eax: grapheme {
+fn read-code-point-utf8-buffered in: (addr buffered-file) -> _/eax: code-point-utf8 {
   var c/eax: byte <- read-byte-buffered in
   var num-trailers/ecx: int <- copy 0
-  $read-grapheme-buffered:compute-length: {
+  $read-code-point-utf8-buffered:compute-length: {
     # single byte: just return it
     compare c, 0xc0
     {
       break-if->=
-      var g/eax: grapheme <- copy c
+      var g/eax: code-point-utf8 <- copy c
       return g
     }
     compare c, 0xfe
     {
       break-if-<
-      var g/eax: grapheme <- copy c
+      var g/eax: code-point-utf8 <- copy c
       return g
     }
     # 2 bytes
@@ -274,23 +274,23 @@ fn read-grapheme-buffered in: (addr buffered-file) -> _/eax: grapheme {
     {
       break-if->=
       num-trailers <- copy 1
-      break $read-grapheme-buffered:compute-length
+      break $read-code-point-utf8-buffered:compute-length
     }
     # 3 bytes
     compare c, 0xf0
     {
       break-if->=
       num-trailers <- copy 2
-      break $read-grapheme-buffered:compute-length
+      break $read-code-point-utf8-buffered:compute-length
     }
     # 4 bytes
     compare c, 0xf8
     {
       break-if->=
       num-trailers <- copy 3
-      break $read-grapheme-buffered:compute-length
+      break $read-code-point-utf8-buffered:compute-length
     }
-$read-grapheme-buffered:abort: {
+$read-code-point-utf8-buffered:abort: {
       # TODO: print to stderr
       print-string-to-real-screen "utf-8 encodings larger than 4 bytes are not supported. First byte seen: "
       var n/eax: int <- copy c
@@ -301,7 +301,7 @@ $read-grapheme-buffered:abort: {
     }
   }
   # prepend trailer bytes
-  var result/edi: grapheme <- copy c
+  var result/edi: code-point-utf8 <- copy c
   var num-byte-shifts/edx: int <- copy 1
   {
     compare num-trailers, 0
@@ -364,23 +364,23 @@ fn test-shift-left-bytes-5 {
   check-ints-equal result, 0, "F - shift-left-bytes >4"
 }
 
-# write a grapheme to a stream of bytes
+# write a code-point-utf8 to a stream of bytes
 # this is like write-to-stream, except we skip leading 0 bytes
-fn write-grapheme out: (addr stream byte), g: grapheme {
-$write-grapheme:body: {
+fn write-code-point-utf8 out: (addr stream byte), g: code-point-utf8 {
+$write-code-point-utf8:body: {
   var c/eax: int <- copy g
   append-byte out, c  # first byte is always written
   c <- shift-right 8
   compare c, 0
-  break-if-= $write-grapheme:body
+  break-if-= $write-code-point-utf8:body
   append-byte out, c
   c <- shift-right 8
   compare c, 0
-  break-if-= $write-grapheme:body
+  break-if-= $write-code-point-utf8:body
   append-byte out, c
   c <- shift-right 8
   compare c, 0
-  break-if-= $write-grapheme:body
+  break-if-= $write-code-point-utf8:body
   append-byte out, c
 }
 }
diff --git a/linux/405screen.mu b/linux/405screen.mu
index c850df2a..ffb0deb8 100644
--- a/linux/405screen.mu
+++ b/linux/405screen.mu
@@ -18,7 +18,7 @@ type screen {
 }
 
 type screen-cell {
-  data: grapheme
+  data: code-point-utf8
   color: int
   background-color: int
   bold?: boolean
@@ -83,7 +83,7 @@ fn clear-screen screen: (addr screen) {
     return
   }
   # fake screen
-  var space/edi: grapheme <- copy 0x20
+  var space/edi: code-point-utf8 <- copy 0x20
   move-cursor screen, 1, 1
   var screen-addr/esi: (addr screen) <- copy screen
   var i/eax: int <- copy 1
@@ -96,7 +96,7 @@ fn clear-screen screen: (addr screen) {
     {
       compare j, *ncols
       break-if->
-      print-grapheme screen, space
+      print-code-point-utf8 screen, space
       j <- increment
       loop
     }
@@ -186,8 +186,8 @@ fn print-stream _screen: (addr screen), s: (addr stream byte) {
     var done?/eax: boolean <- stream-empty? s
     compare done?, 0
     break-if-!=
-    var g/eax: grapheme <- read-grapheme s
-    print-grapheme screen, g
+    var g/eax: code-point-utf8 <- read-code-point-utf8 s
+    print-code-point-utf8 screen, g
     loop
   }
 }
@@ -211,11 +211,11 @@ fn print-array-of-ints-in-decimal screen: (addr screen), _a: (addr array int) {
   }
 }
 
-fn print-grapheme screen: (addr screen), c: grapheme {
+fn print-code-point-utf8 screen: (addr screen), c: code-point-utf8 {
   compare screen, 0
   {
     break-if-!=
-    print-grapheme-to-real-screen c
+    print-code-point-utf8-to-real-screen c
     return
   }
   # fake screen
@@ -239,7 +239,7 @@ fn print-grapheme screen: (addr screen), c: grapheme {
     break-if-<=
     copy-to *cursor-row-addr, num-rows
     # if (top-index > data size) top-index = 0, otherwise top-index += num-cols
-    $print-grapheme:perform-scroll: {
+    $print-code-point-utf8:perform-scroll: {
       var top-index-addr/ebx: (addr int) <- get screen-addr, top-index
       var data-ah/eax: (addr handle array screen-cell) <- get screen-addr, data
       var data/eax: (addr array screen-cell) <- lookup *data-ah
@@ -248,7 +248,7 @@ fn print-grapheme screen: (addr screen), c: grapheme {
       {
         break-if->=
         add-to *top-index-addr, num-cols
-        break $print-grapheme:perform-scroll
+        break $print-code-point-utf8:perform-scroll
       }
       {
         break-if-<
@@ -257,7 +257,7 @@ fn print-grapheme screen: (addr screen), c: grapheme {
     }
   }
   var idx/ecx: int <- current-screen-cell-index screen-addr
-#?   print-string-to-real-screen "printing grapheme at screen index "
+#?   print-string-to-real-screen "printing code-point-utf8 at screen index "
 #?   print-int32-hex-to-real-screen idx
 #?   print-string-to-real-screen ": "
   var data-ah/eax: (addr handle array screen-cell) <- get screen-addr, data
@@ -266,9 +266,9 @@ fn print-grapheme screen: (addr screen), c: grapheme {
   var dest-cell/ecx: (addr screen-cell) <- index data, offset
   var src-cell/eax: (addr screen-cell) <- get screen-addr, curr-attributes
   copy-object src-cell, dest-cell
-  var dest/eax: (addr grapheme) <- get dest-cell, data
-  var c2/ecx: grapheme <- copy c
-#?   print-grapheme-to-real-screen c2
+  var dest/eax: (addr code-point-utf8) <- get dest-cell, data
+  var c2/ecx: code-point-utf8 <- copy c
+#?   print-code-point-utf8-to-real-screen c2
 #?   print-string-to-real-screen "\n"
   copy-to *dest, c2
   increment *cursor-col-addr
@@ -305,21 +305,21 @@ fn screen-cell-index screen-on-stack: (addr screen), row: int, col: int -> _/ecx
   return result
 }
 
-fn screen-grapheme-at screen-on-stack: (addr screen), row: int, col: int -> _/eax: grapheme {
+fn screen-code-point-utf8-at screen-on-stack: (addr screen), row: int, col: int -> _/eax: code-point-utf8 {
   var screen-addr/esi: (addr screen) <- copy screen-on-stack
   var idx/ecx: int <- screen-cell-index screen-addr, row, col
-  var result/eax: grapheme <- screen-grapheme-at-idx screen-addr, idx
+  var result/eax: code-point-utf8 <- screen-code-point-utf8-at-idx screen-addr, idx
   return result
 }
 
-fn screen-grapheme-at-idx screen-on-stack: (addr screen), idx-on-stack: int -> _/eax: grapheme {
+fn screen-code-point-utf8-at-idx screen-on-stack: (addr screen), idx-on-stack: int -> _/eax: code-point-utf8 {
   var screen-addr/esi: (addr screen) <- copy screen-on-stack
   var data-ah/eax: (addr handle array screen-cell) <- get screen-addr, data
   var data/eax: (addr array screen-cell) <- lookup *data-ah
   var idx/ecx: int <- copy idx-on-stack
   var offset/ecx: (offset screen-cell) <- compute-offset data, idx
   var cell/eax: (addr screen-cell) <- index data, offset
-  var src/eax: (addr grapheme) <- get cell, data
+  var src/eax: (addr code-point-utf8) <- get cell, data
   return *src
 }
 
@@ -433,8 +433,8 @@ fn screen-blink-at-idx? screen-on-stack: (addr screen), idx-on-stack: int -> _/e
 }
 
 fn print-code-point screen: (addr screen), c: code-point {
-  var g/eax: grapheme <- to-grapheme c
-  print-grapheme screen, g
+  var g/eax: code-point-utf8 <- to-utf8 c
+  print-code-point-utf8 screen, g
 }
 
 fn print-int32-hex screen: (addr screen), n: int {
@@ -453,8 +453,8 @@ fn print-int32-hex screen: (addr screen), n: int {
     var done?/eax: boolean <- stream-empty? s2-addr
     compare done?, 0
     break-if-!=
-    var g/eax: grapheme <- read-grapheme s2-addr
-    print-grapheme screen, g
+    var g/eax: code-point-utf8 <- read-code-point-utf8 s2-addr
+    print-code-point-utf8 screen, g
     loop
   }
 }
@@ -475,8 +475,8 @@ fn print-int32-hex-bits screen: (addr screen), n: int, bits: int {
     var done?/eax: boolean <- stream-empty? s2-addr
     compare done?, 0
     break-if-!=
-    var g/eax: grapheme <- read-grapheme s2-addr
-    print-grapheme screen, g
+    var g/eax: code-point-utf8 <- read-code-point-utf8 s2-addr
+    print-code-point-utf8 screen, g
     loop
   }
 }
@@ -497,8 +497,8 @@ fn print-int32-decimal screen: (addr screen), n: int {
     var done?/eax: boolean <- stream-empty? s2-addr
     compare done?, 0
     break-if-!=
-    var g/eax: grapheme <- read-grapheme s2-addr
-    print-grapheme screen, g
+    var g/eax: code-point-utf8 <- read-code-point-utf8 s2-addr
+    print-code-point-utf8 screen, g
     loop
   }
 }
@@ -631,7 +631,7 @@ fn check-screen-row screen: (addr screen), row-idx: int, expected: (addr array b
 fn check-screen-row-from screen-on-stack: (addr screen), row-idx: int, col-idx: int, expected: (addr array byte), msg: (addr array byte) {
   var screen/esi: (addr screen) <- copy screen-on-stack
   var idx/ecx: int <- screen-cell-index screen, row-idx, col-idx
-  # compare 'expected' with the screen contents starting at 'idx', grapheme by grapheme
+  # compare 'expected' with the screen contents starting at 'idx', code-point-utf8 by code-point-utf8
   var e: (stream byte 0x100)
   var e-addr/edx: (addr stream byte) <- address e
   write e-addr, expected
@@ -639,35 +639,35 @@ fn check-screen-row-from screen-on-stack: (addr screen), row-idx: int, col-idx:
     var done?/eax: boolean <- stream-empty? e-addr
     compare done?, 0
     break-if-!=
-    var _g/eax: grapheme <- screen-grapheme-at-idx screen, idx
-    var g/ebx: grapheme <- copy _g
-    var expected-grapheme/eax: grapheme <- read-grapheme e-addr
-    # compare graphemes
-    $check-screen-row-from:compare-graphemes: {
-      # if expected-grapheme is space, null grapheme is also ok
+    var _g/eax: code-point-utf8 <- screen-code-point-utf8-at-idx screen, idx
+    var g/ebx: code-point-utf8 <- copy _g
+    var expected-code-point-utf8/eax: code-point-utf8 <- read-code-point-utf8 e-addr
+    # compare code-point-utf8s
+    $check-screen-row-from:compare-code-point-utf8s: {
+      # if expected-code-point-utf8 is space, null code-point-utf8 is also ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         compare g, 0
-        break-if-= $check-screen-row-from:compare-graphemes
+        break-if-= $check-screen-row-from:compare-code-point-utf8s
       }
-      # if (g == expected-grapheme) print "."
-      compare g, expected-grapheme
+      # if (g == expected-code-point-utf8) print "."
+      compare g, expected-code-point-utf8
       {
         break-if-!=
         print-string-to-real-screen "."
-        break $check-screen-row-from:compare-graphemes
+        break $check-screen-row-from:compare-code-point-utf8s
       }
       # otherwise print an error
       print-string-to-real-screen msg
       print-string-to-real-screen ": expected '"
-      print-grapheme-to-real-screen expected-grapheme
+      print-code-point-utf8-to-real-screen expected-code-point-utf8
       print-string-to-real-screen "' at ("
       print-int32-hex-to-real-screen row-idx
       print-string-to-real-screen ", "
       print-int32-hex-to-real-screen col-idx
       print-string-to-real-screen ") but observed '"
-      print-grapheme-to-real-screen g
+      print-code-point-utf8-to-real-screen g
       print-string-to-real-screen "'\n"
     }
     idx <- increment
@@ -685,7 +685,7 @@ fn check-screen-row-in-color screen: (addr screen), fg: int, row-idx: int, expec
 fn check-screen-row-in-color-from screen-on-stack: (addr screen), fg: int, row-idx: int, col-idx: int, expected: (addr array byte), msg: (addr array byte) {
   var screen/esi: (addr screen) <- copy screen-on-stack
   var idx/ecx: int <- screen-cell-index screen, row-idx, col-idx
-  # compare 'expected' with the screen contents starting at 'idx', grapheme by grapheme
+  # compare 'expected' with the screen contents starting at 'idx', code-point-utf8 by code-point-utf8
   var e: (stream byte 0x100)
   var e-addr/edx: (addr stream byte) <- address e
   write e-addr, expected
@@ -693,45 +693,45 @@ fn check-screen-row-in-color-from screen-on-stack: (addr screen), fg: int, row-i
     var done?/eax: boolean <- stream-empty? e-addr
     compare done?, 0
     break-if-!=
-    var _g/eax: grapheme <- screen-grapheme-at-idx screen, idx
-    var g/ebx: grapheme <- copy _g
-    var _expected-grapheme/eax: grapheme <- read-grapheme e-addr
-    var expected-grapheme/edi: grapheme <- copy _expected-grapheme
+    var _g/eax: code-point-utf8 <- screen-code-point-utf8-at-idx screen, idx
+    var g/ebx: code-point-utf8 <- copy _g
+    var _expected-code-point-utf8/eax: code-point-utf8 <- read-code-point-utf8 e-addr
+    var expected-code-point-utf8/edi: code-point-utf8 <- copy _expected-code-point-utf8
     $check-screen-row-in-color-from:compare-cells: {
-      # if expected-grapheme is space, null grapheme is also ok
+      # if expected-code-point-utf8 is space, null code-point-utf8 is also ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         compare g, 0
         break-if-= $check-screen-row-in-color-from:compare-cells
       }
-      # if expected-grapheme is space, a different color is ok
+      # if expected-code-point-utf8 is space, a different color is ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         var color/eax: int <- screen-color-at-idx screen, idx
         compare color, fg
         break-if-!= $check-screen-row-in-color-from:compare-cells
       }
-      # compare graphemes
-      $check-screen-row-in-color-from:compare-graphemes: {
-        # if (g == expected-grapheme) print "."
-        compare g, expected-grapheme
+      # compare code-point-utf8s
+      $check-screen-row-in-color-from:compare-code-point-utf8s: {
+        # if (g == expected-code-point-utf8) print "."
+        compare g, expected-code-point-utf8
         {
           break-if-!=
           print-string-to-real-screen "."
-          break $check-screen-row-in-color-from:compare-graphemes
+          break $check-screen-row-in-color-from:compare-code-point-utf8s
         }
         # otherwise print an error
         print-string-to-real-screen msg
         print-string-to-real-screen ": expected '"
-        print-grapheme-to-real-screen expected-grapheme
+        print-code-point-utf8-to-real-screen expected-code-point-utf8
         print-string-to-real-screen "' at ("
         print-int32-hex-to-real-screen row-idx
         print-string-to-real-screen ", "
         print-int32-hex-to-real-screen col-idx
         print-string-to-real-screen ") but observed '"
-        print-grapheme-to-real-screen g
+        print-code-point-utf8-to-real-screen g
         print-string-to-real-screen "'\n"
       }
       $check-screen-row-in-color-from:compare-colors: {
@@ -745,7 +745,7 @@ fn check-screen-row-in-color-from screen-on-stack: (addr screen), fg: int, row-i
         # otherwise print an error
         print-string-to-real-screen msg
         print-string-to-real-screen ": expected '"
-        print-grapheme-to-real-screen expected-grapheme
+        print-code-point-utf8-to-real-screen expected-code-point-utf8
         print-string-to-real-screen "' at ("
         print-int32-hex-to-real-screen row-idx
         print-string-to-real-screen ", "
@@ -772,7 +772,7 @@ fn check-screen-row-in-background-color screen: (addr screen), bg: int, row-idx:
 fn check-screen-row-in-background-color-from screen-on-stack: (addr screen), bg: int, row-idx: int, col-idx: int, expected: (addr array byte), msg: (addr array byte) {
   var screen/esi: (addr screen) <- copy screen-on-stack
   var idx/ecx: int <- screen-cell-index screen, row-idx, col-idx
-  # compare 'expected' with the screen contents starting at 'idx', grapheme by grapheme
+  # compare 'expected' with the screen contents starting at 'idx', code-point-utf8 by code-point-utf8
   var e: (stream byte 0x100)
   var e-addr/edx: (addr stream byte) <- address e
   write e-addr, expected
@@ -780,45 +780,45 @@ fn check-screen-row-in-background-color-from screen-on-stack: (addr screen), bg:
     var done?/eax: boolean <- stream-empty? e-addr
     compare done?, 0
     break-if-!=
-    var _g/eax: grapheme <- screen-grapheme-at-idx screen, idx
-    var g/ebx: grapheme <- copy _g
-    var _expected-grapheme/eax: grapheme <- read-grapheme e-addr
-    var expected-grapheme/edx: grapheme <- copy _expected-grapheme
+    var _g/eax: code-point-utf8 <- screen-code-point-utf8-at-idx screen, idx
+    var g/ebx: code-point-utf8 <- copy _g
+    var _expected-code-point-utf8/eax: code-point-utf8 <- read-code-point-utf8 e-addr
+    var expected-code-point-utf8/edx: code-point-utf8 <- copy _expected-code-point-utf8
     $check-screen-row-in-background-color-from:compare-cells: {
-      # if expected-grapheme is space, null grapheme is also ok
+      # if expected-code-point-utf8 is space, null code-point-utf8 is also ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         compare g, 0
         break-if-= $check-screen-row-in-background-color-from:compare-cells
       }
-      # if expected-grapheme is space, a different color is ok
+      # if expected-code-point-utf8 is space, a different color is ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         var color/eax: int <- screen-background-color-at-idx screen, idx
         compare color, bg
         break-if-!= $check-screen-row-in-background-color-from:compare-cells
       }
-      # compare graphemes
-      $check-screen-row-in-background-color-from:compare-graphemes: {
-        # if (g == expected-grapheme) print "."
-        compare g, expected-grapheme
+      # compare code-point-utf8s
+      $check-screen-row-in-background-color-from:compare-code-point-utf8s: {
+        # if (g == expected-code-point-utf8) print "."
+        compare g, expected-code-point-utf8
         {
           break-if-!=
           print-string-to-real-screen "."
-          break $check-screen-row-in-background-color-from:compare-graphemes
+          break $check-screen-row-in-background-color-from:compare-code-point-utf8s
         }
         # otherwise print an error
         print-string-to-real-screen msg
         print-string-to-real-screen ": expected '"
-        print-grapheme-to-real-screen expected-grapheme
+        print-code-point-utf8-to-real-screen expected-code-point-utf8
         print-string-to-real-screen "' at ("
         print-int32-hex-to-real-screen row-idx
         print-string-to-real-screen ", "
         print-int32-hex-to-real-screen col-idx
         print-string-to-real-screen ") but observed '"
-        print-grapheme-to-real-screen g
+        print-code-point-utf8-to-real-screen g
         print-string-to-real-screen "'\n"
       }
       $check-screen-row-in-background-color-from:compare-colors: {
@@ -832,7 +832,7 @@ fn check-screen-row-in-background-color-from screen-on-stack: (addr screen), bg:
         # otherwise print an error
         print-string-to-real-screen msg
         print-string-to-real-screen ": expected '"
-        print-grapheme-to-real-screen expected-grapheme
+        print-code-point-utf8-to-real-screen expected-code-point-utf8
         print-string-to-real-screen "' at ("
         print-int32-hex-to-real-screen row-idx
         print-string-to-real-screen ", "
@@ -857,7 +857,7 @@ fn check-screen-row-in-bold screen: (addr screen), row-idx: int, expected: (addr
 fn check-screen-row-in-bold-from screen-on-stack: (addr screen), row-idx: int, col-idx: int, expected: (addr array byte), msg: (addr array byte) {
   var screen/esi: (addr screen) <- copy screen-on-stack
   var idx/ecx: int <- screen-cell-index screen, row-idx, col-idx
-  # compare 'expected' with the screen contents starting at 'idx', grapheme by grapheme
+  # compare 'expected' with the screen contents starting at 'idx', code-point-utf8 by code-point-utf8
   var e: (stream byte 0x100)
   var e-addr/edx: (addr stream byte) <- address e
   write e-addr, expected
@@ -865,45 +865,45 @@ fn check-screen-row-in-bold-from screen-on-stack: (addr screen), row-idx: int, c
     var done?/eax: boolean <- stream-empty? e-addr
     compare done?, 0
     break-if-!=
-    var _g/eax: grapheme <- screen-grapheme-at-idx screen, idx
-    var g/ebx: grapheme <- copy _g
-    var _expected-grapheme/eax: grapheme <- read-grapheme e-addr
-    var expected-grapheme/edx: grapheme <- copy _expected-grapheme
+    var _g/eax: code-point-utf8 <- screen-code-point-utf8-at-idx screen, idx
+    var g/ebx: code-point-utf8 <- copy _g
+    var _expected-code-point-utf8/eax: code-point-utf8 <- read-code-point-utf8 e-addr
+    var expected-code-point-utf8/edx: code-point-utf8 <- copy _expected-code-point-utf8
     $check-screen-row-in-bold-from:compare-cells: {
-      # if expected-grapheme is space, null grapheme is also ok
+      # if expected-code-point-utf8 is space, null code-point-utf8 is also ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         compare g, 0
         break-if-= $check-screen-row-in-bold-from:compare-cells
       }
-      # if expected-grapheme is space, non-bold is ok
+      # if expected-code-point-utf8 is space, non-bold is ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         var bold?/eax: boolean <- screen-bold-at-idx? screen, idx
         compare bold?, 1
         break-if-!= $check-screen-row-in-bold-from:compare-cells
       }
-      # compare graphemes
-      $check-screen-row-in-bold-from:compare-graphemes: {
-        # if (g == expected-grapheme) print "."
-        compare g, expected-grapheme
+      # compare code-point-utf8s
+      $check-screen-row-in-bold-from:compare-code-point-utf8s: {
+        # if (g == expected-code-point-utf8) print "."
+        compare g, expected-code-point-utf8
         {
           break-if-!=
           print-string-to-real-screen "."
-          break $check-screen-row-in-bold-from:compare-graphemes
+          break $check-screen-row-in-bold-from:compare-code-point-utf8s
         }
         # otherwise print an error
         print-string-to-real-screen msg
         print-string-to-real-screen ": expected '"
-        print-grapheme-to-real-screen expected-grapheme
+        print-code-point-utf8-to-real-screen expected-code-point-utf8
         print-string-to-real-screen "' at ("
         print-int32-hex-to-real-screen row-idx
         print-string-to-real-screen ", "
         print-int32-hex-to-real-screen col-idx
         print-string-to-real-screen ") but observed '"
-        print-grapheme-to-real-screen g
+        print-code-point-utf8-to-real-screen g
         print-string-to-real-screen "'\n"
       }
       $check-screen-row-in-bold-from:compare-bold: {
@@ -917,7 +917,7 @@ fn check-screen-row-in-bold-from screen-on-stack: (addr screen), row-idx: int, c
         # otherwise print an error
         print-string-to-real-screen msg
         print-string-to-real-screen ": expected '"
-        print-grapheme-to-real-screen expected-grapheme
+        print-code-point-utf8-to-real-screen expected-code-point-utf8
         print-string-to-real-screen "' at ("
         print-int32-hex-to-real-screen row-idx
         print-string-to-real-screen ", "
@@ -938,7 +938,7 @@ fn check-screen-row-in-underline screen: (addr screen), row-idx: int, expected:
 fn check-screen-row-in-underline-from screen-on-stack: (addr screen), row-idx: int, col-idx: int, expected: (addr array byte), msg: (addr array byte) {
   var screen/esi: (addr screen) <- copy screen-on-stack
   var idx/ecx: int <- screen-cell-index screen, row-idx, col-idx
-  # compare 'expected' with the screen contents starting at 'idx', grapheme by grapheme
+  # compare 'expected' with the screen contents starting at 'idx', code-point-utf8 by code-point-utf8
   var e: (stream byte 0x100)
   var e-addr/edx: (addr stream byte) <- address e
   write e-addr, expected
@@ -946,45 +946,45 @@ fn check-screen-row-in-underline-from screen-on-stack: (addr screen), row-idx: i
     var done?/eax: boolean <- stream-empty? e-addr
     compare done?, 0
     break-if-!=
-    var _g/eax: grapheme <- screen-grapheme-at-idx screen, idx
-    var g/ebx: grapheme <- copy _g
-    var _expected-grapheme/eax: grapheme <- read-grapheme e-addr
-    var expected-grapheme/edx: grapheme <- copy _expected-grapheme
+    var _g/eax: code-point-utf8 <- screen-code-point-utf8-at-idx screen, idx
+    var g/ebx: code-point-utf8 <- copy _g
+    var _expected-code-point-utf8/eax: code-point-utf8 <- read-code-point-utf8 e-addr
+    var expected-code-point-utf8/edx: code-point-utf8 <- copy _expected-code-point-utf8
     $check-screen-row-in-underline-from:compare-cells: {
-      # if expected-grapheme is space, null grapheme is also ok
+      # if expected-code-point-utf8 is space, null code-point-utf8 is also ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         compare g, 0
         break-if-= $check-screen-row-in-underline-from:compare-cells
       }
-      # if expected-grapheme is space, non-underline is ok
+      # if expected-code-point-utf8 is space, non-underline is ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         var underline?/eax: boolean <- screen-underline-at-idx? screen, idx
         compare underline?, 1
         break-if-!= $check-screen-row-in-underline-from:compare-cells
       }
-      # compare graphemes
-      $check-screen-row-in-underline-from:compare-graphemes: {
-        # if (g == expected-grapheme) print "."
-        compare g, expected-grapheme
+      # compare code-point-utf8s
+      $check-screen-row-in-underline-from:compare-code-point-utf8s: {
+        # if (g == expected-code-point-utf8) print "."
+        compare g, expected-code-point-utf8
         {
           break-if-!=
           print-string-to-real-screen "."
-          break $check-screen-row-in-underline-from:compare-graphemes
+          break $check-screen-row-in-underline-from:compare-code-point-utf8s
         }
         # otherwise print an error
         print-string-to-real-screen msg
         print-string-to-real-screen ": expected '"
-        print-grapheme-to-real-screen expected-grapheme
+        print-code-point-utf8-to-real-screen expected-code-point-utf8
         print-string-to-real-screen "' at ("
         print-int32-hex-to-real-screen row-idx
         print-string-to-real-screen ", "
         print-int32-hex-to-real-screen col-idx
         print-string-to-real-screen ") but observed '"
-        print-grapheme-to-real-screen g
+        print-code-point-utf8-to-real-screen g
         print-string-to-real-screen "'\n"
       }
       $check-screen-row-in-underline-from:compare-underline: {
@@ -998,7 +998,7 @@ fn check-screen-row-in-underline-from screen-on-stack: (addr screen), row-idx: i
         # otherwise print an error
         print-string-to-real-screen msg
         print-string-to-real-screen ": expected '"
-        print-grapheme-to-real-screen expected-grapheme
+        print-code-point-utf8-to-real-screen expected-code-point-utf8
         print-string-to-real-screen "' at ("
         print-int32-hex-to-real-screen row-idx
         print-string-to-real-screen ", "
@@ -1019,7 +1019,7 @@ fn check-screen-row-in-reverse screen: (addr screen), row-idx: int, expected: (a
 fn check-screen-row-in-reverse-from screen-on-stack: (addr screen), row-idx: int, col-idx: int, expected: (addr array byte), msg: (addr array byte) {
   var screen/esi: (addr screen) <- copy screen-on-stack
   var idx/ecx: int <- screen-cell-index screen, row-idx, col-idx
-  # compare 'expected' with the screen contents starting at 'idx', grapheme by grapheme
+  # compare 'expected' with the screen contents starting at 'idx', code-point-utf8 by code-point-utf8
   var e: (stream byte 0x100)
   var e-addr/edx: (addr stream byte) <- address e
   write e-addr, expected
@@ -1027,45 +1027,45 @@ fn check-screen-row-in-reverse-from screen-on-stack: (addr screen), row-idx: int
     var done?/eax: boolean <- stream-empty? e-addr
     compare done?, 0
     break-if-!=
-    var _g/eax: grapheme <- screen-grapheme-at-idx screen, idx
-    var g/ebx: grapheme <- copy _g
-    var _expected-grapheme/eax: grapheme <- read-grapheme e-addr
-    var expected-grapheme/edx: grapheme <- copy _expected-grapheme
+    var _g/eax: code-point-utf8 <- screen-code-point-utf8-at-idx screen, idx
+    var g/ebx: code-point-utf8 <- copy _g
+    var _expected-code-point-utf8/eax: code-point-utf8 <- read-code-point-utf8 e-addr
+    var expected-code-point-utf8/edx: code-point-utf8 <- copy _expected-code-point-utf8
     $check-screen-row-in-reverse-from:compare-cells: {
-      # if expected-grapheme is space, null grapheme is also ok
+      # if expected-code-point-utf8 is space, null code-point-utf8 is also ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         compare g, 0
         break-if-= $check-screen-row-in-reverse-from:compare-cells
       }
-      # if expected-grapheme is space, non-reverse is ok
+      # if expected-code-point-utf8 is space, non-reverse is ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         var reverse?/eax: boolean <- screen-reverse-at-idx? screen, idx
         compare reverse?, 1
         break-if-!= $check-screen-row-in-reverse-from:compare-cells
       }
-      # compare graphemes
-      $check-screen-row-in-reverse-from:compare-graphemes: {
-        # if (g == expected-grapheme) print "."
-        compare g, expected-grapheme
+      # compare code-point-utf8s
+      $check-screen-row-in-reverse-from:compare-code-point-utf8s: {
+        # if (g == expected-code-point-utf8) print "."
+        compare g, expected-code-point-utf8
         {
           break-if-!=
           print-string-to-real-screen "."
-          break $check-screen-row-in-reverse-from:compare-graphemes
+          break $check-screen-row-in-reverse-from:compare-code-point-utf8s
         }
         # otherwise print an error
         print-string-to-real-screen msg
         print-string-to-real-screen ": expected '"
-        print-grapheme-to-real-screen expected-grapheme
+        print-code-point-utf8-to-real-screen expected-code-point-utf8
         print-string-to-real-screen "' at ("
         print-int32-hex-to-real-screen row-idx
         print-string-to-real-screen ", "
         print-int32-hex-to-real-screen col-idx
         print-string-to-real-screen ") but observed '"
-        print-grapheme-to-real-screen g
+        print-code-point-utf8-to-real-screen g
         print-string-to-real-screen "'\n"
       }
       $check-screen-row-in-reverse-from:compare-reverse: {
@@ -1079,7 +1079,7 @@ fn check-screen-row-in-reverse-from screen-on-stack: (addr screen), row-idx: int
         # otherwise print an error
         print-string-to-real-screen msg
         print-string-to-real-screen ": expected '"
-        print-grapheme-to-real-screen expected-grapheme
+        print-code-point-utf8-to-real-screen expected-code-point-utf8
         print-string-to-real-screen "' at ("
         print-int32-hex-to-real-screen row-idx
         print-string-to-real-screen ", "
@@ -1100,7 +1100,7 @@ fn check-screen-row-in-blinking screen: (addr screen), row-idx: int, expected: (
 fn check-screen-row-in-blinking-from screen-on-stack: (addr screen), row-idx: int, col-idx: int, expected: (addr array byte), msg: (addr array byte) {
   var screen/esi: (addr screen) <- copy screen-on-stack
   var idx/ecx: int <- screen-cell-index screen, row-idx, col-idx
-  # compare 'expected' with the screen contents starting at 'idx', grapheme by grapheme
+  # compare 'expected' with the screen contents starting at 'idx', code-point-utf8 by code-point-utf8
   var e: (stream byte 0x100)
   var e-addr/edx: (addr stream byte) <- address e
   write e-addr, expected
@@ -1108,45 +1108,45 @@ fn check-screen-row-in-blinking-from screen-on-stack: (addr screen), row-idx: in
     var done?/eax: boolean <- stream-empty? e-addr
     compare done?, 0
     break-if-!=
-    var _g/eax: grapheme <- screen-grapheme-at-idx screen, idx
-    var g/ebx: grapheme <- copy _g
-    var _expected-grapheme/eax: grapheme <- read-grapheme e-addr
-    var expected-grapheme/edx: grapheme <- copy _expected-grapheme
+    var _g/eax: code-point-utf8 <- screen-code-point-utf8-at-idx screen, idx
+    var g/ebx: code-point-utf8 <- copy _g
+    var _expected-code-point-utf8/eax: code-point-utf8 <- read-code-point-utf8 e-addr
+    var expected-code-point-utf8/edx: code-point-utf8 <- copy _expected-code-point-utf8
     $check-screen-row-in-blinking-from:compare-cells: {
-      # if expected-grapheme is space, null grapheme is also ok
+      # if expected-code-point-utf8 is space, null code-point-utf8 is also ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         compare g, 0
         break-if-= $check-screen-row-in-blinking-from:compare-cells
       }
-      # if expected-grapheme is space, non-blinking is ok
+      # if expected-code-point-utf8 is space, non-blinking is ok
       {
-        compare expected-grapheme, 0x20
+        compare expected-code-point-utf8, 0x20
         break-if-!=
         var blinking?/eax: boolean <- screen-blink-at-idx? screen, idx
         compare blinking?, 1
         break-if-!= $check-screen-row-in-blinking-from:compare-cells
       }
-      # compare graphemes
-      $check-screen-row-in-blinking-from:compare-graphemes: {
-        # if (g == expected-grapheme) print "."
-        compare g, expected-grapheme
+      # compare code-point-utf8s
+      $check-screen-row-in-blinking-from:compare-code-point-utf8s: {
+        # if (g == expected-code-point-utf8) print "."
+        compare g, expected-code-point-utf8
         {
           break-if-!=
           print-string-to-real-screen "."
-          break $check-screen-row-in-blinking-from:compare-graphemes
+          break $check-screen-row-in-blinking-from:compare-code-point-utf8s
         }
         # otherwise print an error
         print-string-to-real-screen msg
         print-string-to-real-screen ": expected '"
-        print-grapheme-to-real-screen expected-grapheme
+        print-code-point-utf8-to-real-screen expected-code-point-utf8
         print-string-to-real-screen "' at ("
         print-int32-hex-to-real-screen row-idx
         print-string-to-real-screen ", "
         print-int32-hex-to-real-screen col-idx
         print-string-to-real-screen ") but observed '"
-        print-grapheme-to-real-screen g
+        print-code-point-utf8-to-real-screen g
         print-string-to-real-screen "'\n"
       }
       $check-screen-row-in-blinking-from:compare-blinking: {
@@ -1160,7 +1160,7 @@ fn check-screen-row-in-blinking-from screen-on-stack: (addr screen), row-idx: in
         # otherwise print an error
         print-string-to-real-screen msg
         print-string-to-real-screen ": expected '"
-        print-grapheme-to-real-screen expected-grapheme
+        print-code-point-utf8-to-real-screen expected-code-point-utf8
         print-string-to-real-screen "' at ("
         print-int32-hex-to-real-screen row-idx
         print-string-to-real-screen ", "
@@ -1175,21 +1175,21 @@ fn check-screen-row-in-blinking-from screen-on-stack: (addr screen), row-idx: in
   }
 }
 
-fn test-print-single-grapheme {
+fn test-print-single-code-point-utf8 {
   var screen-on-stack: screen
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
-  check-screen-row screen, 1/row, "a", "F - test-print-single-grapheme"  # top-left corner of the screen
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
+  check-screen-row screen, 1/row, "a", "F - test-print-single-code-point-utf8"  # top-left corner of the screen
 }
 
-fn test-print-multiple-graphemes {
+fn test-print-multiple-code-point-utf8s {
   var screen-on-stack: screen
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
   print-string screen, "Hello, 世界"
-  check-screen-row screen, 1/row, "Hello, 世界", "F - test-print-multiple-graphemes"
+  check-screen-row screen, 1/row, "Hello, 世界", "F - test-print-multiple-code-point-utf8s"
 }
 
 fn test-move-cursor {
@@ -1197,8 +1197,8 @@ fn test-move-cursor {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
   move-cursor screen, 1, 4
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   check-screen-row screen, 1/row, "   a", "F - test-move-cursor"  # top row
 }
 
@@ -1207,8 +1207,8 @@ fn test-move-cursor-zeroes {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
   move-cursor screen, 0, 0
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   check-screen-row screen, 1/row, "a", "F - test-move-cursor-zeroes"  # top-left corner of the screen
 }
 
@@ -1217,8 +1217,8 @@ fn test-move-cursor-zero-row {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
   move-cursor screen, 0, 2
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   check-screen-row screen, 1/row, " a", "F - test-move-cursor-zero-row"  # top row
 }
 
@@ -1227,8 +1227,8 @@ fn test-move-cursor-zero-column {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
   move-cursor screen, 4, 0
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   check-screen-row screen, 4/row, "a", "F - test-move-cursor-zero-column"
 }
 
@@ -1237,8 +1237,8 @@ fn test-move-cursor-negative-row {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5, 3
   move-cursor screen, -1/row, 2/col
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   # no move
   check-screen-row screen, 1/row, "a", "F - test-move-cursor-negative-row"
 }
@@ -1248,8 +1248,8 @@ fn test-move-cursor-negative-column {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5, 3
   move-cursor screen, 2/row, -1/col
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   # no move
   check-screen-row screen, 1/row, "a", "F - test-move-cursor-negative-column"
 }
@@ -1259,8 +1259,8 @@ fn test-move-cursor-column-too-large {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 3/cols
   move-cursor screen, 1/row, 4/col
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   # top row is empty
   check-screen-row screen, 1/row, "   ", "F - test-move-cursor-column-too-large"
   # character shows up on next row
@@ -1272,8 +1272,8 @@ fn test-move-cursor-column-too-large-saturates {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 3/cols
   move-cursor screen, 1/row, 6/col
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   # top row is empty
   check-screen-row screen, 1/row, "   ", "F - test-move-cursor-column-too-large-saturates"  # top-left corner of the screen
   # character shows up at the start of next row
@@ -1285,8 +1285,8 @@ fn test-move-cursor-row-too-large {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 3/cols
   move-cursor screen, 6/row, 2/col
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   # bottom row shows the character
   check-screen-row screen, 5/row, " a", "F - test-move-cursor-row-too-large"
 }
@@ -1296,8 +1296,8 @@ fn test-move-cursor-row-too-large-saturates {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 3/cols
   move-cursor screen, 9/row, 2/col
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   # bottom row shows the character
   check-screen-row screen, 5/row, " a", "F - test-move-cursor-row-too-large-saturates"
 }
@@ -1307,8 +1307,8 @@ fn test-check-screen-row-from {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
   move-cursor screen, 1, 4
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   check-screen-row screen, 1/row, "   a", "F - test-check-screen-row-from/baseline"
   check-screen-row-from screen, 1/row, 4/col, "a", "F - test-check-screen-row-from"
 }
@@ -1328,8 +1328,8 @@ fn test-check-screen-scrolls-on-overflow {
   initialize-screen screen, 5/rows, 4/cols
   # single character starting at bottom right
   move-cursor screen, 5/rows, 4/cols
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   check-screen-row-from screen, 5/row, 4/col, "a", "F - test-check-screen-scrolls-on-overflow/baseline"  # bottom-right corner of the screen
   # multiple characters starting at bottom right
   move-cursor screen, 5, 4
@@ -1348,14 +1348,14 @@ fn test-check-screen-color {
   var screen-on-stack: screen
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   start-color screen, 1/fg, 0/bg
   c <- copy 0x62/b
-  print-grapheme screen, c
+  print-code-point-utf8 screen, c
   start-color screen, 0/fg, 7/bg
   c <- copy 0x63/c
-  print-grapheme screen, c
+  print-code-point-utf8 screen, c
   check-screen-row-in-color screen, 0/fg, 1/row, "a c", "F - test-check-screen-color"
 }
 
@@ -1363,14 +1363,14 @@ fn test-check-screen-background-color {
   var screen-on-stack: screen
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   start-color screen, 0/fg, 1/bg
   c <- copy 0x62/b
-  print-grapheme screen, c
+  print-code-point-utf8 screen, c
   start-color screen, 0/fg, 7/bg
   c <- copy 0x63/c
-  print-grapheme screen, c
+  print-code-point-utf8 screen, c
   check-screen-row-in-background-color screen, 7/bg, 1/row, "a c", "F - test-check-screen-background-color"
 }
 
@@ -1379,14 +1379,14 @@ fn test-check-screen-bold {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
   start-bold screen
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   reset-formatting screen
   c <- copy 0x62/b
-  print-grapheme screen, c
+  print-code-point-utf8 screen, c
   start-bold screen
   c <- copy 0x63/c
-  print-grapheme screen, c
+  print-code-point-utf8 screen, c
   check-screen-row-in-bold screen, 1/row, "a c", "F - test-check-screen-bold"
 }
 
@@ -1395,14 +1395,14 @@ fn test-check-screen-underline {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
   start-underline screen
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   reset-formatting screen
   c <- copy 0x62/b
-  print-grapheme screen, c
+  print-code-point-utf8 screen, c
   start-underline screen
   c <- copy 0x63/c
-  print-grapheme screen, c
+  print-code-point-utf8 screen, c
   check-screen-row-in-underline screen, 1/row, "a c", "F - test-check-screen-underline"
 }
 
@@ -1411,14 +1411,14 @@ fn test-check-screen-reverse {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
   start-reverse-video screen
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   reset-formatting screen
   c <- copy 0x62/b
-  print-grapheme screen, c
+  print-code-point-utf8 screen, c
   start-reverse-video screen
   c <- copy 0x63/c
-  print-grapheme screen, c
+  print-code-point-utf8 screen, c
   check-screen-row-in-reverse screen, 1/row, "a c", "F - test-check-screen-reverse"
 }
 
@@ -1427,14 +1427,14 @@ fn test-check-screen-blinking {
   var screen/esi: (addr screen) <- address screen-on-stack
   initialize-screen screen, 5/rows, 4/cols
   start-blinking screen
-  var c/eax: grapheme <- copy 0x61/a
-  print-grapheme screen, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  print-code-point-utf8 screen, c
   reset-formatting screen
   c <- copy 0x62/b
-  print-grapheme screen, c
+  print-code-point-utf8 screen, c
   start-blinking screen
   c <- copy 0x63/c
-  print-grapheme screen, c
+  print-code-point-utf8 screen, c
   check-screen-row-in-blinking screen, 1/row, "a c", "F - test-check-screen-blinking"
 }
 
diff --git a/linux/407right-justify.mu b/linux/407right-justify.mu
index b7322ae5..aa767782 100644
--- a/linux/407right-justify.mu
+++ b/linux/407right-justify.mu
@@ -6,7 +6,7 @@ fn print-int32-decimal-right-justified screen: (addr screen), n: int, _width: in
   {
     compare n-width, width
     break-if->=
-    print-grapheme screen, 0x20/space
+    print-code-point-utf8 screen, 0x20/space
     width <- decrement
     loop
   }
diff --git a/linux/411string.mu b/linux/411string.mu
index cf0471ac..493c9b56 100644
--- a/linux/411string.mu
+++ b/linux/411string.mu
@@ -1,4 +1,4 @@
-# read up to 'len' graphemes after skipping the first 'start' ones
+# read up to 'len' code-point-utf8s after skipping the first 'start' ones
 fn substring in: (addr array byte), start: int, len: int, out-ah: (addr handle array byte) {
   var in-stream: (stream byte 0x100)
   var in-stream-addr/esi: (addr stream byte) <- address in-stream
@@ -6,29 +6,29 @@ fn substring in: (addr array byte), start: int, len: int, out-ah: (addr handle a
   var out-stream: (stream byte 0x100)
   var out-stream-addr/edi: (addr stream byte) <- address out-stream
   $substring:core: {
-    # skip 'start' graphemes
+    # skip 'start' code-point-utf8s
     var i/eax: int <- copy 0
     {
       compare i, start
       break-if->=
       {
-        var dummy/eax: grapheme <- read-grapheme in-stream-addr
+        var dummy/eax: code-point-utf8 <- read-code-point-utf8 in-stream-addr
         compare dummy, 0xffffffff/end-of-file
         break-if-= $substring:core
       }
       i <- increment
       loop
     }
-    # copy 'len' graphemes
+    # copy 'len' code-point-utf8s
     i <- copy 0
     {
       compare i, len
       break-if->=
       {
-        var g/eax: grapheme <- read-grapheme in-stream-addr
+        var g/eax: code-point-utf8 <- read-code-point-utf8 in-stream-addr
         compare g, 0xffffffff/end-of-file
         break-if-= $substring:core
-        write-grapheme out-stream-addr, g
+        write-code-point-utf8 out-stream-addr, g
       }
       i <- increment
       loop
@@ -85,7 +85,7 @@ fn test-substring {
   check-strings-equal out, "bcde", "F - test-substring/middle-too-small"
 }
 
-fn split-string in: (addr array byte), delim: grapheme, out: (addr handle array (handle array byte)) {
+fn split-string in: (addr array byte), delim: code-point-utf8, out: (addr handle array (handle array byte)) {
   var in-stream: (stream byte 0x100)
   var in-stream-addr/esi: (addr stream byte) <- address in-stream
   write in-stream-addr, in
@@ -94,10 +94,10 @@ fn split-string in: (addr array byte), delim: grapheme, out: (addr handle array
   var curr-stream: (stream byte 0x100)
   var curr-stream-addr/ecx: (addr stream byte) <- address curr-stream
   $split-string:core: {
-    var g/eax: grapheme <- read-grapheme in-stream-addr
+    var g/eax: code-point-utf8 <- read-code-point-utf8 in-stream-addr
     compare g, 0xffffffff
     break-if-=
-#?     print-grapheme-to-real-screen g
+#?     print-code-point-utf8-to-real-screen g
 #?     print-string-to-real-screen "\n"
     compare g, delim
     {
@@ -110,7 +110,7 @@ fn split-string in: (addr array byte), delim: grapheme, out: (addr handle array
       clear-stream curr-stream-addr
       loop $split-string:core
     }
-    write-grapheme curr-stream-addr, g
+    write-code-point-utf8 curr-stream-addr, g
     loop
   }
   stream-to-array tokens-stream-addr, out
diff --git a/linux/apps/arith.mu b/linux/apps/arith.mu
index 4393a34c..08b2008d 100644
--- a/linux/apps/arith.mu
+++ b/linux/apps/arith.mu
@@ -33,7 +33,7 @@
 
 fn main -> _/ebx: int {
   enable-keyboard-immediate-mode
-  var look/esi: grapheme <- copy 0  # lookahead
+  var look/esi: code-point-utf8 <- copy 0  # lookahead
   var n/eax: int <- copy 0  # result of each expression
   print-string 0/screen, "press ctrl-c or ctrl-d to exit\n"
   # read-eval-print loop
@@ -55,17 +55,17 @@ fn main -> _/ebx: int {
   return 0
 }
 
-fn simplify -> _/eax: int, _/esi: grapheme {
+fn simplify -> _/eax: int, _/esi: code-point-utf8 {
   # prime the pump
-  var look/esi: grapheme <- get-char
+  var look/esi: code-point-utf8 <- get-char
   # do it
   var result/eax: int <- copy 0
   result, look <- expression look
   return result, look
 }
 
-fn expression _look: grapheme -> _/eax: int, _/esi: grapheme {
-  var look/esi: grapheme <- copy _look
+fn expression _look: code-point-utf8 -> _/eax: int, _/esi: code-point-utf8 {
+  var look/esi: code-point-utf8 <- copy _look
   # read arg
   var result/eax: int <- copy 0
   result, look <- term look
@@ -78,7 +78,7 @@ fn expression _look: grapheme -> _/eax: int, _/esi: grapheme {
       break-if-= $expression:loop
     }
     # read operator
-    var op/ecx: grapheme <- copy 0
+    var op/ecx: code-point-utf8 <- copy 0
     op, look <- operator look
     # read next arg
     var second/edx: int <- copy 0
@@ -109,8 +109,8 @@ fn expression _look: grapheme -> _/eax: int, _/esi: grapheme {
   return result, look
 }
 
-fn term _look: grapheme -> _/eax: int, _/esi: grapheme {
-  var look/esi: grapheme <- copy _look
+fn term _look: code-point-utf8 -> _/eax: int, _/esi: code-point-utf8 {
+  var look/esi: code-point-utf8 <- copy _look
   # read arg
   look <- skip-spaces look
   var result/eax: int <- copy 0
@@ -124,7 +124,7 @@ fn term _look: grapheme -> _/eax: int, _/esi: grapheme {
       break-if-= $term:loop
     }
     # read operator
-    var op/ecx: grapheme <- copy 0
+    var op/ecx: code-point-utf8 <- copy 0
     op, look <- operator look
     # read next arg
     var second/edx: int <- copy 0
@@ -154,8 +154,8 @@ fn term _look: grapheme -> _/eax: int, _/esi: grapheme {
   return result, look
 }
 
-fn factor _look: grapheme -> _/eax: int, _/esi: grapheme {
-  var look/esi: grapheme <- copy _look  # should be a no-op
+fn factor _look: code-point-utf8 -> _/eax: int, _/esi: code-point-utf8 {
+  var look/esi: code-point-utf8 <- copy _look  # should be a no-op
   look <- skip-spaces look
   # if next char is not '(', parse a number
   compare look, 0x28/open-paren
@@ -174,7 +174,7 @@ fn factor _look: grapheme -> _/eax: int, _/esi: grapheme {
   return result, look
 }
 
-fn mul-or-div? c: grapheme -> _/eax: boolean {
+fn mul-or-div? c: code-point-utf8 -> _/eax: boolean {
   compare c, 0x2a/*
   {
     break-if-!=
@@ -188,7 +188,7 @@ fn mul-or-div? c: grapheme -> _/eax: boolean {
   return 0/false
 }
 
-fn add-or-sub? c: grapheme -> _/eax: boolean {
+fn add-or-sub? c: code-point-utf8 -> _/eax: boolean {
   compare c, 0x2b/+
   {
     break-if-!=
@@ -202,14 +202,14 @@ fn add-or-sub? c: grapheme -> _/eax: boolean {
   return 0/false
 }
 
-fn operator _look: grapheme -> _/ecx: grapheme, _/esi: grapheme {
-  var op/ecx: grapheme <- copy _look
-  var look/esi: grapheme <- get-char
+fn operator _look: code-point-utf8 -> _/ecx: code-point-utf8, _/esi: code-point-utf8 {
+  var op/ecx: code-point-utf8 <- copy _look
+  var look/esi: code-point-utf8 <- get-char
   return op, look
 }
 
-fn num _look: grapheme -> _/eax: int, _/esi: grapheme {
-  var look/esi: grapheme <- copy _look
+fn num _look: code-point-utf8 -> _/eax: int, _/esi: code-point-utf8 {
+  var look/esi: code-point-utf8 <- copy _look
   var result/edi: int <- copy 0
   {
     var first-digit/eax: int <- to-decimal-digit look
@@ -234,8 +234,8 @@ fn num _look: grapheme -> _/eax: int, _/esi: grapheme {
   return result, look
 }
 
-fn skip-spaces _look: grapheme -> _/esi: grapheme {
-  var look/esi: grapheme <- copy _look  # should be a no-op
+fn skip-spaces _look: code-point-utf8 -> _/esi: code-point-utf8 {
+  var look/esi: code-point-utf8 <- copy _look  # should be a no-op
   {
     compare look, 0x20
     break-if-!=
@@ -245,9 +245,9 @@ fn skip-spaces _look: grapheme -> _/esi: grapheme {
   return look
 }
 
-fn get-char -> _/esi: grapheme {
-  var look/eax: grapheme <- read-key-from-real-keyboard
-  print-grapheme-to-real-screen look
+fn get-char -> _/esi: code-point-utf8 {
+  var look/eax: code-point-utf8 <- read-key-from-real-keyboard
+  print-code-point-utf8-to-real-screen look
   compare look, 4
   {
     break-if-!=
diff --git a/linux/apps/parse-int.mu b/linux/apps/parse-int.mu
index 0f8c71d1..ccff8d44 100644
--- a/linux/apps/parse-int.mu
+++ b/linux/apps/parse-int.mu
@@ -37,7 +37,7 @@ fn parse-int _in: (addr array byte) -> _/eax: int {
     var tmp/ebx: (addr byte) <- index in, i
     var c/eax: byte <- copy-byte *tmp
     #
-    var g/eax: grapheme <- copy c
+    var g/eax: code-point-utf8 <- copy c
     var digit/eax: int <- to-decimal-digit g
     result <- add digit
     i <- increment
diff --git a/linux/apps/print-file.mu b/linux/apps/print-file.mu
index 75ce2e39..284b805e 100644
--- a/linux/apps/print-file.mu
+++ b/linux/apps/print-file.mu
@@ -30,8 +30,8 @@ fn main _args: (addr array addr array byte) -> _/ebx: int {
       var c/eax: byte <- read-byte-buffered in-addr
       compare c, 0xffffffff/end-of-file
       break-if-=
-      var g/eax: grapheme <- copy c
-      print-grapheme 0/screen, g
+      var g/eax: code-point-utf8 <- copy c
+      print-code-point-utf8 0/screen, g
       loop
     }
   }
diff --git a/linux/apps/tui.mu b/linux/apps/tui.mu
index 4e58b986..f4fc914c 100644
--- a/linux/apps/tui.mu
+++ b/linux/apps/tui.mu
@@ -23,7 +23,7 @@ fn main -> _/ebx: int {
 
   print-string 0/screen, "press a key to see its code: "
   enable-keyboard-immediate-mode
-  var x/eax: grapheme <- read-key-from-real-keyboard
+  var x/eax: code-point-utf8 <- read-key-from-real-keyboard
   enable-keyboard-type-mode
   enable-screen-type-mode
   print-string 0/screen, "You pressed "
diff --git a/linux/browse/main.mu b/linux/browse/main.mu
index 5b4f2e06..27504afe 100644
--- a/linux/browse/main.mu
+++ b/linux/browse/main.mu
@@ -49,7 +49,7 @@ fn interactive fs: (addr buffered-file) {
   #
   {
     render paginated-screen, fs
-    var key/eax: grapheme <- read-key-from-real-keyboard
+    var key/eax: code-point-utf8 <- read-key-from-real-keyboard
     compare key, 0x71/'q'
     loop-if-!=
   }
@@ -160,13 +160,13 @@ fn test-render-asterisk-in-text {
 fn render-normal screen: (addr paginated-screen), fs: (addr buffered-file) {
   var newline-seen?/esi: boolean <- copy 0/false
   var start-of-paragraph?/edi: boolean <- copy 1/true
-  var previous-grapheme/ebx: grapheme <- copy 0
+  var previous-code-point-utf8/ebx: code-point-utf8 <- copy 0
 $render-normal:loop: {
     # if done-drawing?(screen) break
     var done?/eax: boolean <- done-drawing? screen
     compare done?, 0/false
     break-if-!=
-    var c/eax: grapheme <- read-grapheme-buffered fs
+    var c/eax: code-point-utf8 <- read-code-point-utf8-buffered fs
 $render-normal:loop-body: {
       # if (c == EOF) break
       compare c, 0xffffffff/end-of-file
@@ -186,8 +186,8 @@ $render-normal:loop-body: {
         # otherwise render two newlines
         {
           break-if-=
-          add-grapheme screen, 0xa/newline
-          add-grapheme screen, 0xa/newline
+          add-code-point-utf8 screen, 0xa/newline
+          add-code-point-utf8 screen, 0xa/newline
           newline-seen? <- copy 0/false
           start-of-paragraph? <- copy 1/true
           break $render-normal:loop-body
@@ -221,20 +221,20 @@ $render-normal:flush-buffered-newline: {
         {
           compare c, 0x20
           break-if-!=
-          add-grapheme screen, 0xa/newline
+          add-code-point-utf8 screen, 0xa/newline
           break $render-normal:flush-buffered-newline
         }
-        add-grapheme screen, 0x20/space
+        add-code-point-utf8 screen, 0x20/space
         # fall through to print c
       }
       ## end soft newline support
 
 $render-normal:whitespace-separated-regions: {
-        # if previous-grapheme wasn't whitespace, skip this block
+        # if previous-code-point-utf8 wasn't whitespace, skip this block
         {
-          compare previous-grapheme, 0x20/space
+          compare previous-code-point-utf8, 0x20/space
           break-if-=
-          compare previous-grapheme, 0xa/newline
+          compare previous-code-point-utf8, 0xa/newline
           break-if-=
           break $render-normal:whitespace-separated-regions
         }
@@ -260,9 +260,9 @@ $render-normal:whitespace-separated-regions: {
         }
       }
       #
-      add-grapheme screen, c
+      add-code-point-utf8 screen, c
     }  # $render-normal:loop-body
-    previous-grapheme <- copy c
+    previous-code-point-utf8 <- copy c
     loop
   }  # $render-normal:loop
 }
@@ -271,7 +271,7 @@ fn render-header-line screen: (addr paginated-screen), fs: (addr buffered-file)
 $render-header-line:body: {
   # compute color based on number of '#'s
   var header-level/esi: int <- copy 1  # caller already grabbed one
-  var c/eax: grapheme <- copy 0
+  var c/eax: code-point-utf8 <- copy 0
   {
     # if done-drawing?(screen) return
     {
@@ -280,7 +280,7 @@ $render-header-line:body: {
       break-if-!= $render-header-line:body
     }
     #
-    c <- read-grapheme-buffered fs
+    c <- read-code-point-utf8-buffered fs
     # if (c != '#') break
     compare c, 0x23/'#'
     break-if-!=
@@ -298,7 +298,7 @@ $render-header-line:body: {
       break-if-!=
     }
     #
-    c <- read-grapheme-buffered fs
+    c <- read-code-point-utf8-buffered fs
     # if (c == EOF) break
     compare c, 0xffffffff/end-of-file
     break-if-=
@@ -306,7 +306,7 @@ $render-header-line:body: {
     compare c, 0xa/newline
     break-if-=
     #
-    add-grapheme screen, c
+    add-code-point-utf8 screen, c
     #
     loop
   }
@@ -353,7 +353,7 @@ fn render-until-asterisk screen: (addr paginated-screen), fs: (addr buffered-fil
     compare done?, 0/false
     break-if-!=
     #
-    var c/eax: grapheme <- read-grapheme-buffered fs
+    var c/eax: code-point-utf8 <- read-code-point-utf8-buffered fs
     # if (c == EOF) break
     compare c, 0xffffffff/end-of-file
     break-if-=
@@ -361,7 +361,7 @@ fn render-until-asterisk screen: (addr paginated-screen), fs: (addr buffered-fil
     compare c, 0x2a/'*'
     break-if-=
     #
-    add-grapheme screen, c
+    add-code-point-utf8 screen, c
     #
     loop
   }
@@ -374,7 +374,7 @@ fn render-until-underscore screen: (addr paginated-screen), fs: (addr buffered-f
     compare done?, 0/false
     break-if-!=
     #
-    var c/eax: grapheme <- read-grapheme-buffered fs
+    var c/eax: code-point-utf8 <- read-code-point-utf8-buffered fs
     # if (c == EOF) break
     compare c, 0xffffffff/end-of-file
     break-if-=
@@ -382,7 +382,7 @@ fn render-until-underscore screen: (addr paginated-screen), fs: (addr buffered-f
     compare c, 0x5f/'_'
     break-if-=
     #
-    add-grapheme screen, c
+    add-code-point-utf8 screen, c
     #
     loop
   }
diff --git a/linux/browse/paginated-screen.mu b/linux/browse/paginated-screen.mu
index f4579d95..05f954a7 100644
--- a/linux/browse/paginated-screen.mu
+++ b/linux/browse/paginated-screen.mu
@@ -7,7 +7,7 @@
 #   on each frame
 #     start-drawing
 #     while !done-drawing
-#       add-grapheme ...
+#       add-code-point-utf8 ...
 
 type paginated-screen {
   screen: (handle screen)
@@ -152,23 +152,23 @@ fn done-drawing? _self: (addr paginated-screen) -> _/eax: boolean {
   return 1/true
 }
 
-fn add-grapheme _self: (addr paginated-screen), c: grapheme {
-#?   print-string-to-real-screen "add-grapheme: "
-#?   print-grapheme-to-real-screen c
+fn add-code-point-utf8 _self: (addr paginated-screen), c: code-point-utf8 {
+#?   print-string-to-real-screen "add-code-point-utf8: "
+#?   print-code-point-utf8-to-real-screen c
 #?   print-string-to-real-screen "\n"
-$add-grapheme:body: {
+$add-code-point-utf8:body: {
   var self/esi: (addr paginated-screen) <- copy _self
   {
     compare c, 0xa/newline
     break-if-!=
     next-line self
     reposition-cursor self
-    break $add-grapheme:body
+    break $add-code-point-utf8:body
   }
   # print c
   var screen-ah/eax: (addr handle screen) <- get self, screen
   var screen-addr/eax: (addr screen) <- lookup *screen-ah
-  print-grapheme screen-addr, c
+  print-code-point-utf8 screen-addr, c
   # self->col++
   var tmp/eax: (addr int) <- get self, col
   increment *tmp
@@ -186,21 +186,21 @@ $add-grapheme:body: {
 
 ## tests
 
-fn test-print-grapheme-on-paginated-screen {
+fn test-print-code-point-utf8-on-paginated-screen {
   var pg-on-stack: paginated-screen
   var pg/eax: (addr paginated-screen) <- address pg-on-stack
   initialize-fake-paginated-screen pg, 3/rows, 0xa/cols, 0xa/page-width, 0, 0
   start-drawing pg
   {
-    var c/ecx: grapheme <- copy 0x61/a
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x61/a
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
-    check-ints-equal done, 0, "F - test-print-grapheme-on-paginated-screen/done"
+    check-ints-equal done, 0, "F - test-print-code-point-utf8-on-paginated-screen/done"
   }
   var screen-ah/eax: (addr handle screen) <- get pg, screen
   var screen-addr/eax: (addr screen) <- lookup *screen-ah
-  check-screen-row screen-addr, 1, "a", "F - test-print-grapheme-on-paginated-screen"
+  check-screen-row screen-addr, 1, "a", "F - test-print-code-point-utf8-on-paginated-screen"
 }
 
 fn test-print-single-page {
@@ -210,29 +210,29 @@ fn test-print-single-page {
   start-drawing pg
   # pages at columns [1, 3), [3, 5)
   {
-    var c/ecx: grapheme <- copy 0x61/a
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x61/a
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page/done-1"
   }
   {
-    var c/ecx: grapheme <- copy 0x62/b
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x62/b
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page/done-2"
   }
   {
-    var c/ecx: grapheme <- copy 0x63/c
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x63/c
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page/done-3"
   }
   {
-    var c/ecx: grapheme <- copy 0x64/d
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x64/d
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page/done-4"
@@ -250,36 +250,36 @@ fn test-print-single-page-narrower-than-page-width {
   initialize-fake-paginated-screen pg, 2/rows, 4/cols, 5/page-width, 0, 0
   start-drawing pg
   {
-    var c/ecx: grapheme <- copy 0x61/a
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x61/a
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page-narrower-than-page-width/done-1"
   }
   {
-    var c/ecx: grapheme <- copy 0x62/b
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x62/b
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page-narrower-than-page-width/done-2"
   }
   {
-    var c/ecx: grapheme <- copy 0x63/c
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x63/c
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page-narrower-than-page-width/done-3"
   }
   {
-    var c/ecx: grapheme <- copy 0x64/d
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x64/d
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page-narrower-than-page-width/done-4"
   }
   {
-    var c/ecx: grapheme <- copy 0x65/e
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x65/e
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page-narrower-than-page-width/done-5"
@@ -297,36 +297,36 @@ fn test-print-single-page-narrower-than-page-width-with-margin {
   initialize-fake-paginated-screen pg, 2/rows, 4/cols, 5/page-width, 0/top-margin, 1/left-margin
   start-drawing pg
   {
-    var c/ecx: grapheme <- copy 0x61/a
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x61/a
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page-narrower-than-page-width-with-margin/done-1"
   }
   {
-    var c/ecx: grapheme <- copy 0x62/b
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x62/b
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page-narrower-than-page-width-with-margin/done-2"
   }
   {
-    var c/ecx: grapheme <- copy 0x63/c
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x63/c
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page-narrower-than-page-width-with-margin/done-3"
   }
   {
-    var c/ecx: grapheme <- copy 0x64/d
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x64/d
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page-narrower-than-page-width-with-margin/done-4"
   }
   {
-    var c/ecx: grapheme <- copy 0x65/e
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x65/e
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-single-page-narrower-than-page-width-with-margin/done-5"
@@ -344,29 +344,29 @@ fn test-print-multiple-pages {
   initialize-fake-paginated-screen pg, 2/rows, 2/cols, 1/page-width, 0, 0
   start-drawing pg
   {
-    var c/ecx: grapheme <- copy 0x61/a
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x61/a
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-multiple-pages/done-1"
   }
   {
-    var c/ecx: grapheme <- copy 0x62/b
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x62/b
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-multiple-pages/done-2"
   }
   {
-    var c/ecx: grapheme <- copy 0x63/c
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x63/c
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-multiple-pages/done-3"
   }
   {
-    var c/ecx: grapheme <- copy 0x64/d
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x64/d
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 1, "F - test-print-multiple-pages/done-4"
@@ -384,57 +384,57 @@ fn test-print-multiple-pages-2 {
   initialize-fake-paginated-screen pg, 2/rows, 4/cols, 2/page-width, 0, 0
   start-drawing pg
   {
-    var c/ecx: grapheme <- copy 0x61/a
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x61/a
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-multiple-pages-2/done-1"
   }
   {
-    var c/ecx: grapheme <- copy 0x62/b
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x62/b
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-multiple-pages-2/done-2"
   }
   {
-    var c/ecx: grapheme <- copy 0x63/c
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x63/c
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-multiple-pages-2/done-3"
   }
   {
-    var c/ecx: grapheme <- copy 0x64/d
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x64/d
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-multiple-pages-2/done-4"
   }
   {
-    var c/ecx: grapheme <- copy 0x65/e
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x65/e
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-multiple-pages-2/done-5"
   }
   {
-    var c/ecx: grapheme <- copy 0x66/f
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x66/f
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-multiple-pages-2/done-6"
   }
   {
-    var c/ecx: grapheme <- copy 0x67/g
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x67/g
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 0, "F - test-print-multiple-pages-2/done-7"
   }
   {
-    var c/ecx: grapheme <- copy 0x68/h
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x68/h
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
     check-ints-equal done, 1, "F - test-print-multiple-pages-2/done-8"
@@ -452,60 +452,60 @@ fn test-print-multiple-pages-with-margins {
   initialize-fake-paginated-screen pg, 3/rows, 6/cols, 2/page-width, 1/top-margin, 1/left-margin
   start-drawing pg
   {
-    var c/ecx: grapheme <- copy 0x61/a
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x61/a
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
-    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/grapheme-1"
+    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/code-point-utf8-1"
   }
   {
-    var c/ecx: grapheme <- copy 0x62/b
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x62/b
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
-    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/grapheme-2"
+    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/code-point-utf8-2"
   }
   {
-    var c/ecx: grapheme <- copy 0x63/c
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x63/c
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
-    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/grapheme-3"
+    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/code-point-utf8-3"
   }
   {
-    var c/ecx: grapheme <- copy 0x64/d
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x64/d
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
-    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/grapheme-4"
+    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/code-point-utf8-4"
   }
   {
-    var c/ecx: grapheme <- copy 0x65/e
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x65/e
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
-    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/grapheme-5"
+    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/code-point-utf8-5"
   }
   {
-    var c/ecx: grapheme <- copy 0x66/f
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x66/f
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
-    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/grapheme-6"
+    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/code-point-utf8-6"
   }
   {
-    var c/ecx: grapheme <- copy 0x67/g
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x67/g
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
-    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/grapheme-7"
+    check-ints-equal done, 0, "F - test-print-multiple-pages-with-margins/code-point-utf8-7"
   }
   {
-    var c/ecx: grapheme <- copy 0x68/h
-    add-grapheme pg, c
+    var c/ecx: code-point-utf8 <- copy 0x68/h
+    add-code-point-utf8 pg, c
     var done?/eax: boolean <- done-drawing? pg
     var done/eax: int <- copy done?
-    check-ints-equal done, 1, "F - test-print-multiple-pages-with-margins/grapheme-8"
+    check-ints-equal done, 1, "F - test-print-multiple-pages-with-margins/code-point-utf8-8"
   }
   var screen-ah/eax: (addr handle screen) <- get pg, screen
   var screen-addr/eax: (addr screen) <- lookup *screen-ah
diff --git a/linux/mu b/linux/mu
index fdeacea0..94cb0a6e 100755
Binary files a/linux/mu and b/linux/mu differ
diff --git a/linux/mu.subx b/linux/mu.subx
index 879b751e..b4c55006 100644
--- a/linux/mu.subx
+++ b/linux/mu.subx
@@ -416,8 +416,8 @@ Type-id:  # (stream (addr array byte))
   "stream"/imm32  # 11
   "slice"/imm32  # 12
   "code-point"/imm32  # 13; smallest scannable unit from a text stream
-  "grapheme"/imm32  # 14; smallest printable unit; will eventually be composed of multiple code-points, but currently corresponds 1:1
-                    # only 4-byte graphemes in utf-8 are currently supported;
+  "code-point-utf8"/imm32  # 14; smallest printable unit; will eventually be composed of multiple code-points, but currently corresponds 1:1
+                    # only 4-byte code-point-utf8s in utf-8 are currently supported;
                     # unclear how we should deal with larger clusters.
   "float"/imm32     # 15
   # 0x40
@@ -22183,9 +22183,9 @@ $mu-numberlike-output?:check-code-point:
     (simple-mu-type? %esi 0xd)  # code-point => eax
     3d/compare-eax-and 0/imm32/false
     75/jump-if-!= $mu-numberlike-output?:return-true/disp8
-$mu-numberlike-output?:check-grapheme:
-    # if t is a grapheme, return
-    (simple-mu-type? %esi 0xe)  # grapheme => eax
+$mu-numberlike-output?:check-code-point-utf8:
+    # if t is a code-point-utf8, return
+    (simple-mu-type? %esi 0xe)  # code-point-utf8 => eax
     3d/compare-eax-and 0/imm32/false
     75/jump-if-!= $mu-numberlike-output?:return-true/disp8
 $mu-numberlike-output?:return-false:
diff --git a/linux/tile/box.mu b/linux/tile/box.mu
index 859d0b8e..f5155ea8 100644
--- a/linux/tile/box.mu
+++ b/linux/tile/box.mu
@@ -78,7 +78,7 @@ fn clear-rect screen: (addr screen), row1: int, col1: int, row2: int, col2: int
     {
       compare j, col2
       break-if->
-      print-grapheme screen 0x20/space
+      print-code-point-utf8 screen 0x20/space
       j <- increment
       loop
     }
@@ -98,7 +98,7 @@ fn clear-rect2 screen: (addr screen), row1: int, col1: int, w: int, h: int {
     {
       compare j, h
       break-if->=
-      print-grapheme screen 0x20/space
+      print-code-point-utf8 screen 0x20/space
       j <- increment
       loop
     }
diff --git a/linux/tile/environment.mu b/linux/tile/environment.mu
index 3c869d3c..1512db6d 100644
--- a/linux/tile/environment.mu
+++ b/linux/tile/environment.mu
@@ -70,7 +70,7 @@ fn initialize-environment-with-fake-screen _self: (addr environment), nrows: int
 # Iterate
 #############
 
-fn process _self: (addr environment), key: grapheme {
+fn process _self: (addr environment), key: code-point-utf8 {
   var self/esi: (addr environment) <- copy _self
   var fn-name-ah/eax: (addr handle word) <- get self, partial-function-name
   var fn-name/eax: (addr word) <- lookup *fn-name-ah
@@ -102,7 +102,7 @@ fn process _self: (addr environment), key: grapheme {
 }
 
 # collect new name in partial-function-name, and move the cursor to function with that name
-fn process-goto-dialog _self: (addr environment), key: grapheme {
+fn process-goto-dialog _self: (addr environment), key: code-point-utf8 {
   var self/esi: (addr environment) <- copy _self
   var fn-name-ah/edi: (addr handle word) <- get self, partial-function-name
   # if 'esc' pressed, cancel goto
@@ -130,7 +130,7 @@ fn process-goto-dialog _self: (addr environment), key: grapheme {
   compare key, 0x7f/del  # backspace on Macs
   $process-goto-dialog:backspace: {
     break-if-!=
-    # if not at start, delete grapheme before cursor
+    # if not at start, delete code-point-utf8 before cursor
     var fn-name/eax: (addr word) <- lookup *fn-name-ah
     var at-start?/eax: boolean <- cursor-at-start? fn-name
     compare at-start?, 0/false
@@ -142,24 +142,24 @@ fn process-goto-dialog _self: (addr environment), key: grapheme {
     return
   }
   # otherwise insert key within current word
-  var print?/eax: boolean <- real-grapheme? key
-  $process-goto-dialog:real-grapheme: {
+  var print?/eax: boolean <- real-code-point-utf8? key
+  $process-goto-dialog:real-code-point-utf8: {
     compare print?, 0/false
     break-if-=
     var fn-name/eax: (addr word) <- lookup *fn-name-ah
-    add-grapheme-to-word fn-name, key
+    add-code-point-utf8-to-word fn-name, key
     return
   }
   # silently ignore other hotkeys
 }
 
-fn process-function _self: (addr environment), _function: (addr function), key: grapheme {
+fn process-function _self: (addr environment), _function: (addr function), key: code-point-utf8 {
   var self/esi: (addr environment) <- copy _self
   var function/edi: (addr function) <- copy _function
   process-function-edit self, function, key
 }
 
-fn process-function-edit _self: (addr environment), _function: (addr function), key: grapheme {
+fn process-function-edit _self: (addr environment), _function: (addr function), key: code-point-utf8 {
   var self/esi: (addr environment) <- copy _self
   var function/edi: (addr function) <- copy _function
   var cursor-word-ah/ebx: (addr handle word) <- get function, cursor-word
@@ -290,7 +290,7 @@ fn process-function-edit _self: (addr environment), _function: (addr function),
   compare key, 0x7f/del  # backspace on Macs
   $process-function-edit:backspace: {
     break-if-!=
-    # if not at start of some word, delete grapheme before cursor within current word
+    # if not at start of some word, delete code-point-utf8 before cursor within current word
     var at-start?/eax: boolean <- cursor-at-start? cursor-word
     compare at-start?, 0/false
     {
@@ -325,25 +325,25 @@ fn process-function-edit _self: (addr environment), _function: (addr function),
       copy-object new-prev-word-ah, cursor-word-ah
       return
     }
-    # if start of word is quote and grapheme before cursor is not, just insert it as usual
+    # if start of word is quote and code-point-utf8 before cursor is not, just insert it as usual
     # TODO: support string escaping
     {
-      var first-grapheme/eax: grapheme <- first-grapheme cursor-word
-      compare first-grapheme, 0x22/double-quote
+      var first-code-point-utf8/eax: code-point-utf8 <- first-code-point-utf8 cursor-word
+      compare first-code-point-utf8, 0x22/double-quote
       break-if-!=
-      var final-grapheme/eax: grapheme <- grapheme-before-cursor cursor-word
-      compare final-grapheme, 0x22/double-quote
+      var final-code-point-utf8/eax: code-point-utf8 <- code-point-utf8-before-cursor cursor-word
+      compare final-code-point-utf8, 0x22/double-quote
       break-if-=
       break $process-function-edit:space
     }
-    # if start of word is '[' and grapheme before cursor is not ']', just insert it as usual
+    # if start of word is '[' and code-point-utf8 before cursor is not ']', just insert it as usual
     # TODO: support nested arrays
     {
-      var first-grapheme/eax: grapheme <- first-grapheme cursor-word
-      compare first-grapheme, 0x5b/[
+      var first-code-point-utf8/eax: code-point-utf8 <- first-code-point-utf8 cursor-word
+      compare first-code-point-utf8, 0x5b/[
       break-if-!=
-      var final-grapheme/eax: grapheme <- grapheme-before-cursor cursor-word
-      compare final-grapheme, 0x5d/]
+      var final-code-point-utf8/eax: code-point-utf8 <- code-point-utf8-before-cursor cursor-word
+      compare final-code-point-utf8, 0x5d/]
       break-if-=
       break $process-function-edit:space
     }
@@ -368,26 +368,26 @@ fn process-function-edit _self: (addr environment), _function: (addr function),
       var at-end?/eax: boolean <- cursor-at-end? cursor-word
       compare at-end?, 0/false
       break-if-!=
-      var g/eax: grapheme <- pop-after-cursor cursor-word
-      add-grapheme-to-word next-word, g
+      var g/eax: code-point-utf8 <- pop-after-cursor cursor-word
+      add-code-point-utf8-to-word next-word, g
       loop
     }
     cursor-to-start next-word
     return
   }
   # otherwise insert key within current word
-  var g/edx: grapheme <- copy key
-  var print?/eax: boolean <- real-grapheme? key
-  $process-function-edit:real-grapheme: {
+  var g/edx: code-point-utf8 <- copy key
+  var print?/eax: boolean <- real-code-point-utf8? key
+  $process-function-edit:real-code-point-utf8: {
     compare print?, 0/false
     break-if-=
-    add-grapheme-to-word cursor-word, g
+    add-code-point-utf8-to-word cursor-word, g
     return
   }
   # silently ignore other hotkeys
 }
 
-fn process-sandbox _self: (addr environment), _sandbox: (addr sandbox), key: grapheme {
+fn process-sandbox _self: (addr environment), _sandbox: (addr sandbox), key: code-point-utf8 {
   var self/esi: (addr environment) <- copy _self
   var sandbox/edi: (addr sandbox) <- copy _sandbox
   var rename-word-mode-ah?/ecx: (addr handle word) <- get sandbox, partial-name-for-cursor-word
@@ -413,7 +413,7 @@ fn process-sandbox _self: (addr environment), _sandbox: (addr sandbox), key: gra
   process-sandbox-edit self, sandbox, key
 }
 
-fn process-sandbox-edit _self: (addr environment), _sandbox: (addr sandbox), key: grapheme {
+fn process-sandbox-edit _self: (addr environment), _sandbox: (addr sandbox), key: code-point-utf8 {
   var self/esi: (addr environment) <- copy _self
   var sandbox/edi: (addr sandbox) <- copy _sandbox
   var cursor-call-path-ah/eax: (addr handle call-path-element) <- get sandbox, cursor-call-path
@@ -730,7 +730,7 @@ fn process-sandbox-edit _self: (addr environment), _sandbox: (addr sandbox), key
   compare key, 0x7f/del  # backspace on Macs
   $process-sandbox-edit:backspace: {
     break-if-!=
-    # if not at start of some word, delete grapheme before cursor within current word
+    # if not at start of some word, delete code-point-utf8 before cursor within current word
     var at-start?/eax: boolean <- cursor-at-start? cursor-word
     compare at-start?, 0/false
     {
@@ -766,25 +766,25 @@ fn process-sandbox-edit _self: (addr environment), _sandbox: (addr sandbox), key
       decrement-final-element cursor-call-path
       return
     }
-    # if start of word is quote and grapheme before cursor is not, just insert it as usual
+    # if start of word is quote and code-point-utf8 before cursor is not, just insert it as usual
     # TODO: support string escaping
     {
-      var first-grapheme/eax: grapheme <- first-grapheme cursor-word
-      compare first-grapheme, 0x22/double-quote
+      var first-code-point-utf8/eax: code-point-utf8 <- first-code-point-utf8 cursor-word
+      compare first-code-point-utf8, 0x22/double-quote
       break-if-!=
-      var final-grapheme/eax: grapheme <- grapheme-before-cursor cursor-word
-      compare final-grapheme, 0x22/double-quote
+      var final-code-point-utf8/eax: code-point-utf8 <- code-point-utf8-before-cursor cursor-word
+      compare final-code-point-utf8, 0x22/double-quote
       break-if-=
       break $process-sandbox-edit:space
     }
-    # if start of word is '[' and grapheme before cursor is not ']', just insert it as usual
+    # if start of word is '[' and code-point-utf8 before cursor is not ']', just insert it as usual
     # TODO: support nested arrays
     {
-      var first-grapheme/eax: grapheme <- first-grapheme cursor-word
-      compare first-grapheme, 0x5b/[
+      var first-code-point-utf8/eax: code-point-utf8 <- first-code-point-utf8 cursor-word
+      compare first-code-point-utf8, 0x5b/[
       break-if-!=
-      var final-grapheme/eax: grapheme <- grapheme-before-cursor cursor-word
-      compare final-grapheme, 0x5d/]
+      var final-code-point-utf8/eax: code-point-utf8 <- code-point-utf8-before-cursor cursor-word
+      compare final-code-point-utf8, 0x5d/]
       break-if-=
       break $process-sandbox-edit:space
     }
@@ -809,8 +809,8 @@ fn process-sandbox-edit _self: (addr environment), _sandbox: (addr sandbox), key
       var at-end?/eax: boolean <- cursor-at-end? cursor-word
       compare at-end?, 0/false
       break-if-!=
-      var g/eax: grapheme <- pop-after-cursor cursor-word
-      add-grapheme-to-word next-word, g
+      var g/eax: code-point-utf8 <- pop-after-cursor cursor-word
+      add-code-point-utf8-to-word next-word, g
       loop
     }
     cursor-to-start next-word
@@ -838,12 +838,12 @@ fn process-sandbox-edit _self: (addr environment), _sandbox: (addr sandbox), key
     return
   }
   # otherwise insert key within current word
-  var g/edx: grapheme <- copy key
-  var print?/eax: boolean <- real-grapheme? key
-  $process-sandbox-edit:real-grapheme: {
+  var g/edx: code-point-utf8 <- copy key
+  var print?/eax: boolean <- real-code-point-utf8? key
+  $process-sandbox-edit:real-code-point-utf8: {
     compare print?, 0/false
     break-if-=
-    add-grapheme-to-word cursor-word, g
+    add-code-point-utf8-to-word cursor-word, g
     return
   }
   # silently ignore other hotkeys
@@ -852,7 +852,7 @@ fn process-sandbox-edit _self: (addr environment), _sandbox: (addr sandbox), key
 # collect new name in partial-name-for-cursor-word, and then rename the word
 # at cursor to it
 # Precondition: cursor-call-path is a singleton (not within a call)
-fn process-sandbox-rename _sandbox: (addr sandbox), key: grapheme {
+fn process-sandbox-rename _sandbox: (addr sandbox), key: code-point-utf8 {
   var sandbox/esi: (addr sandbox) <- copy _sandbox
   var new-name-ah/edi: (addr handle word) <- get sandbox, partial-name-for-cursor-word
   # if 'esc' pressed, cancel rename
@@ -911,7 +911,7 @@ fn process-sandbox-rename _sandbox: (addr sandbox), key: grapheme {
     {
       var new-name/eax: (addr word) <- lookup *new-name-ah
       cursor-to-start new-name
-      add-grapheme-to-word new-name, 0x3d/=
+      add-code-point-utf8-to-word new-name, 0x3d/=
     }
     # append name to new line
     chain-words new-line-word-ah, new-name-ah
@@ -941,7 +941,7 @@ fn process-sandbox-rename _sandbox: (addr sandbox), key: grapheme {
   compare key, 0x7f/del  # backspace on Macs
   $process-sandbox-rename:backspace: {
     break-if-!=
-    # if not at start, delete grapheme before cursor
+    # if not at start, delete code-point-utf8 before cursor
     var new-name/eax: (addr word) <- lookup *new-name-ah
     var at-start?/eax: boolean <- cursor-at-start? new-name
     compare at-start?, 0/false
@@ -953,12 +953,12 @@ fn process-sandbox-rename _sandbox: (addr sandbox), key: grapheme {
     return
   }
   # otherwise insert key within current word
-  var print?/eax: boolean <- real-grapheme? key
-  $process-sandbox-rename:real-grapheme: {
+  var print?/eax: boolean <- real-code-point-utf8? key
+  $process-sandbox-rename:real-code-point-utf8: {
     compare print?, 0/false
     break-if-=
     var new-name/eax: (addr word) <- lookup *new-name-ah
-    add-grapheme-to-word new-name, key
+    add-code-point-utf8-to-word new-name, key
     return
   }
   # silently ignore other hotkeys
@@ -968,7 +968,7 @@ fn process-sandbox-rename _sandbox: (addr sandbox), key: grapheme {
 # of the sandbox to be a new function with that name. Replace the last line
 # with a call to the appropriate function.
 # Precondition: cursor-call-path is a singleton (not within a call)
-fn process-sandbox-define _sandbox: (addr sandbox), functions: (addr handle function), key: grapheme {
+fn process-sandbox-define _sandbox: (addr sandbox), functions: (addr handle function), key: code-point-utf8 {
   var sandbox/esi: (addr sandbox) <- copy _sandbox
   var new-name-ah/edi: (addr handle word) <- get sandbox, partial-name-for-function
   # if 'esc' pressed, cancel define
@@ -1033,7 +1033,7 @@ fn process-sandbox-define _sandbox: (addr sandbox), functions: (addr handle func
   compare key, 0x7f/del  # backspace on Macs
   $process-sandbox-define:backspace: {
     break-if-!=
-    # if not at start, delete grapheme before cursor
+    # if not at start, delete code-point-utf8 before cursor
     var new-name/eax: (addr word) <- lookup *new-name-ah
     var at-start?/eax: boolean <- cursor-at-start? new-name
     compare at-start?, 0/false
@@ -1045,12 +1045,12 @@ fn process-sandbox-define _sandbox: (addr sandbox), functions: (addr handle func
     return
   }
   # otherwise insert key within current word
-  var print?/eax: boolean <- real-grapheme? key
-  $process-sandbox-define:real-grapheme: {
+  var print?/eax: boolean <- real-code-point-utf8? key
+  $process-sandbox-define:real-code-point-utf8: {
     compare print?, 0/false
     break-if-=
     var new-name/eax: (addr word) <- lookup *new-name-ah
-    add-grapheme-to-word new-name, key
+    add-code-point-utf8-to-word new-name, key
     return
   }
   # silently ignore other hotkeys
@@ -2107,7 +2107,7 @@ fn render-function-right-aligned screen: (addr screen), row: int, right-col: int
   start-color screen, 0, 0xf7
   clear-rect screen, row, col, new-row, col2
   col <- add 1
-#?   var dummy/eax: grapheme <- read-key-from-real-keyboard
+#?   var dummy/eax: code-point-utf8 <- read-key-from-real-keyboard
   render-function screen, row, col, f
   new-row <- add 1/function-bottom-margin
   col <- subtract 1/function-left-padding
@@ -2144,7 +2144,7 @@ fn render-function screen: (addr screen), row: int, col: int, _f: (addr function
   render-line-without-stack screen, body, row, col, cursor-word, cursor-row, cursor-col
 }
 
-fn real-grapheme? g: grapheme -> _/eax: boolean {
+fn real-code-point-utf8? g: code-point-utf8 -> _/eax: boolean {
   # if g == newline return true
   compare g, 0xa
   {
diff --git a/linux/tile/gap-buffer.mu b/linux/tile/gap-buffer.mu
index 0132daf0..1441684b 100644
--- a/linux/tile/gap-buffer.mu
+++ b/linux/tile/gap-buffer.mu
@@ -1,14 +1,14 @@
 type gap-buffer {
-  left: grapheme-stack
-  right: grapheme-stack
+  left: code-point-utf8-stack
+  right: code-point-utf8-stack
 }
 
 fn initialize-gap-buffer _self: (addr gap-buffer) {
   var self/esi: (addr gap-buffer) <- copy _self
-  var left/eax: (addr grapheme-stack) <- get self, left
-  initialize-grapheme-stack left, 0x10/max-word-size
-  var right/eax: (addr grapheme-stack) <- get self, right
-  initialize-grapheme-stack right, 0x10/max-word-size
+  var left/eax: (addr code-point-utf8-stack) <- get self, left
+  initialize-code-point-utf8-stack left, 0x10/max-word-size
+  var right/eax: (addr code-point-utf8-stack) <- get self, right
+  initialize-code-point-utf8-stack right, 0x10/max-word-size
 }
 
 # just for tests
@@ -21,8 +21,8 @@ fn initialize-gap-buffer-with self: (addr gap-buffer), s: (addr array byte) {
     var done?/eax: boolean <- stream-empty? stream
     compare done?, 0/false
     break-if-!=
-    var g/eax: grapheme <- read-grapheme stream
-    add-grapheme-at-gap self, g
+    var g/eax: code-point-utf8 <- read-code-point-utf8 stream
+    add-code-point-utf8-at-gap self, g
     loop
   }
 }
@@ -37,44 +37,44 @@ fn gap-buffer-to-string self: (addr gap-buffer), out: (addr handle array byte) {
 fn emit-gap-buffer _self: (addr gap-buffer), out: (addr stream byte) {
   var self/esi: (addr gap-buffer) <- copy _self
   clear-stream out
-  var left/eax: (addr grapheme-stack) <- get self, left
+  var left/eax: (addr code-point-utf8-stack) <- get self, left
   emit-stack-from-bottom left, out
-  var right/eax: (addr grapheme-stack) <- get self, right
+  var right/eax: (addr code-point-utf8-stack) <- get self, right
   emit-stack-from-top right, out
 }
 
 # dump stack from bottom to top
-fn emit-stack-from-bottom _self: (addr grapheme-stack), out: (addr stream byte) {
-  var self/esi: (addr grapheme-stack) <- copy _self
-  var data-ah/edi: (addr handle array grapheme) <- get self, data
-  var _data/eax: (addr array grapheme) <- lookup *data-ah
-  var data/edi: (addr array grapheme) <- copy _data
+fn emit-stack-from-bottom _self: (addr code-point-utf8-stack), out: (addr stream byte) {
+  var self/esi: (addr code-point-utf8-stack) <- copy _self
+  var data-ah/edi: (addr handle array code-point-utf8) <- get self, data
+  var _data/eax: (addr array code-point-utf8) <- lookup *data-ah
+  var data/edi: (addr array code-point-utf8) <- copy _data
   var top-addr/ecx: (addr int) <- get self, top
   var i/eax: int <- copy 0
   {
     compare i, *top-addr
     break-if->=
-    var g/edx: (addr grapheme) <- index data, i
-    write-grapheme out, *g
+    var g/edx: (addr code-point-utf8) <- index data, i
+    write-code-point-utf8 out, *g
     i <- increment
     loop
   }
 }
 
 # dump stack from top to bottom
-fn emit-stack-from-top _self: (addr grapheme-stack), out: (addr stream byte) {
-  var self/esi: (addr grapheme-stack) <- copy _self
-  var data-ah/edi: (addr handle array grapheme) <- get self, data
-  var _data/eax: (addr array grapheme) <- lookup *data-ah
-  var data/edi: (addr array grapheme) <- copy _data
+fn emit-stack-from-top _self: (addr code-point-utf8-stack), out: (addr stream byte) {
+  var self/esi: (addr code-point-utf8-stack) <- copy _self
+  var data-ah/edi: (addr handle array code-point-utf8) <- get self, data
+  var _data/eax: (addr array code-point-utf8) <- lookup *data-ah
+  var data/edi: (addr array code-point-utf8) <- copy _data
   var top-addr/ecx: (addr int) <- get self, top
   var i/eax: int <- copy *top-addr
   i <- decrement
   {
     compare i, 0
     break-if-<
-    var g/edx: (addr grapheme) <- index data, i
-    write-grapheme out, *g
+    var g/edx: (addr code-point-utf8) <- index data, i
+    write-code-point-utf8 out, *g
     i <- decrement
     loop
   }
@@ -82,33 +82,33 @@ fn emit-stack-from-top _self: (addr grapheme-stack), out: (addr stream byte) {
 
 fn render-gap-buffer screen: (addr screen), _gap: (addr gap-buffer) {
   var gap/esi: (addr gap-buffer) <- copy _gap
-  var left/eax: (addr grapheme-stack) <- get gap, left
+  var left/eax: (addr code-point-utf8-stack) <- get gap, left
   render-stack-from-bottom left, screen
-  var right/eax: (addr grapheme-stack) <- get gap, right
+  var right/eax: (addr code-point-utf8-stack) <- get gap, right
   render-stack-from-top right, screen
 }
 
 fn gap-buffer-length _gap: (addr gap-buffer) -> _/eax: int {
   var gap/esi: (addr gap-buffer) <- copy _gap
-  var left/eax: (addr grapheme-stack) <- get gap, left
+  var left/eax: (addr code-point-utf8-stack) <- get gap, left
   var tmp/eax: (addr int) <- get left, top
   var left-length/ecx: int <- copy *tmp
-  var right/esi: (addr grapheme-stack) <- get gap, right
+  var right/esi: (addr code-point-utf8-stack) <- get gap, right
   tmp <- get right, top
   var result/eax: int <- copy *tmp
   result <- add left-length
   return result
 }
 
-fn add-grapheme-at-gap _self: (addr gap-buffer), g: grapheme {
+fn add-code-point-utf8-at-gap _self: (addr gap-buffer), g: code-point-utf8 {
   var self/esi: (addr gap-buffer) <- copy _self
-  var left/eax: (addr grapheme-stack) <- get self, left
-  push-grapheme-stack left, g
+  var left/eax: (addr code-point-utf8-stack) <- get self, left
+  push-code-point-utf8-stack left, g
 }
 
 fn gap-to-start self: (addr gap-buffer) {
   {
-    var curr/eax: grapheme <- gap-left self
+    var curr/eax: code-point-utf8 <- gap-left self
     compare curr, -1
     loop-if-!=
   }
@@ -116,7 +116,7 @@ fn gap-to-start self: (addr gap-buffer) {
 
 fn gap-to-end self: (addr gap-buffer) {
   {
-    var curr/eax: grapheme <- gap-right self
+    var curr/eax: code-point-utf8 <- gap-right self
     compare curr, -1
     loop-if-!=
   }
@@ -124,96 +124,96 @@ fn gap-to-end self: (addr gap-buffer) {
 
 fn gap-at-start? _self: (addr gap-buffer) -> _/eax: boolean {
   var self/esi: (addr gap-buffer) <- copy _self
-  var left/eax: (addr grapheme-stack) <- get self, left
-  var result/eax: boolean <- grapheme-stack-empty? left
+  var left/eax: (addr code-point-utf8-stack) <- get self, left
+  var result/eax: boolean <- code-point-utf8-stack-empty? left
   return result
 }
 
 fn gap-at-end? _self: (addr gap-buffer) -> _/eax: boolean {
   var self/esi: (addr gap-buffer) <- copy _self
-  var right/eax: (addr grapheme-stack) <- get self, right
-  var result/eax: boolean <- grapheme-stack-empty? right
+  var right/eax: (addr code-point-utf8-stack) <- get self, right
+  var result/eax: boolean <- code-point-utf8-stack-empty? right
   return result
 }
 
-fn gap-right _self: (addr gap-buffer) -> _/eax: grapheme {
+fn gap-right _self: (addr gap-buffer) -> _/eax: code-point-utf8 {
   var self/esi: (addr gap-buffer) <- copy _self
-  var g/eax: grapheme <- copy 0
-  var right/ecx: (addr grapheme-stack) <- get self, right
-  g <- pop-grapheme-stack right
+  var g/eax: code-point-utf8 <- copy 0
+  var right/ecx: (addr code-point-utf8-stack) <- get self, right
+  g <- pop-code-point-utf8-stack right
   compare g, -1
   {
     break-if-=
-    var left/ecx: (addr grapheme-stack) <- get self, left
-    push-grapheme-stack left, g
+    var left/ecx: (addr code-point-utf8-stack) <- get self, left
+    push-code-point-utf8-stack left, g
   }
   return g
 }
 
-fn gap-left _self: (addr gap-buffer) -> _/eax: grapheme {
+fn gap-left _self: (addr gap-buffer) -> _/eax: code-point-utf8 {
   var self/esi: (addr gap-buffer) <- copy _self
-  var g/eax: grapheme <- copy 0
+  var g/eax: code-point-utf8 <- copy 0
   {
-    var left/ecx: (addr grapheme-stack) <- get self, left
-    g <- pop-grapheme-stack left
+    var left/ecx: (addr code-point-utf8-stack) <- get self, left
+    g <- pop-code-point-utf8-stack left
   }
   compare g, -1
   {
     break-if-=
-    var right/ecx: (addr grapheme-stack) <- get self, right
-    push-grapheme-stack right, g
+    var right/ecx: (addr code-point-utf8-stack) <- get self, right
+    push-code-point-utf8-stack right, g
   }
   return g
 }
 
 fn gap-index _self: (addr gap-buffer) -> _/eax: int {
   var self/eax: (addr gap-buffer) <- copy _self
-  var left/eax: (addr grapheme-stack) <- get self, left
+  var left/eax: (addr code-point-utf8-stack) <- get self, left
   var top-addr/eax: (addr int) <- get left, top
   var result/eax: int <- copy *top-addr
   return result
 }
 
-fn first-grapheme-in-gap-buffer _self: (addr gap-buffer) -> _/eax: grapheme {
+fn first-code-point-utf8-in-gap-buffer _self: (addr gap-buffer) -> _/eax: code-point-utf8 {
   var self/esi: (addr gap-buffer) <- copy _self
   # try to read from left
-  var left/eax: (addr grapheme-stack) <- get self, left
+  var left/eax: (addr code-point-utf8-stack) <- get self, left
   var top-addr/ecx: (addr int) <- get left, top
   compare *top-addr, 0
   {
     break-if-<=
-    var data-ah/eax: (addr handle array grapheme) <- get left, data
-    var data/eax: (addr array grapheme) <- lookup *data-ah
-    var result-addr/eax: (addr grapheme) <- index data, 0
+    var data-ah/eax: (addr handle array code-point-utf8) <- get left, data
+    var data/eax: (addr array code-point-utf8) <- lookup *data-ah
+    var result-addr/eax: (addr code-point-utf8) <- index data, 0
     return *result-addr
   }
   # try to read from right
-  var right/eax: (addr grapheme-stack) <- get self, right
+  var right/eax: (addr code-point-utf8-stack) <- get self, right
   top-addr <- get right, top
   compare *top-addr, 0
   {
     break-if-<=
-    var data-ah/eax: (addr handle array grapheme) <- get right, data
-    var data/eax: (addr array grapheme) <- lookup *data-ah
+    var data-ah/eax: (addr handle array code-point-utf8) <- get right, data
+    var data/eax: (addr array code-point-utf8) <- lookup *data-ah
     var top/ecx: int <- copy *top-addr
     top <- decrement
-    var result-addr/eax: (addr grapheme) <- index data, top
+    var result-addr/eax: (addr code-point-utf8) <- index data, top
     return *result-addr
   }
   # give up
   return -1
 }
 
-fn grapheme-before-cursor-in-gap-buffer _self: (addr gap-buffer) -> _/eax: grapheme {
+fn code-point-utf8-before-cursor-in-gap-buffer _self: (addr gap-buffer) -> _/eax: code-point-utf8 {
   var self/esi: (addr gap-buffer) <- copy _self
   # try to read from left
-  var left/ecx: (addr grapheme-stack) <- get self, left
+  var left/ecx: (addr code-point-utf8-stack) <- get self, left
   var top-addr/edx: (addr int) <- get left, top
   compare *top-addr, 0
   {
     break-if-<=
-    var result/eax: grapheme <- pop-grapheme-stack left
-    push-grapheme-stack left, result
+    var result/eax: code-point-utf8 <- pop-code-point-utf8-stack left
+    push-code-point-utf8-stack left, result
     return result
   }
   # give up
@@ -222,27 +222,27 @@ fn grapheme-before-cursor-in-gap-buffer _self: (addr gap-buffer) -> _/eax: graph
 
 fn delete-before-gap _self: (addr gap-buffer) {
   var self/eax: (addr gap-buffer) <- copy _self
-  var left/eax: (addr grapheme-stack) <- get self, left
-  var dummy/eax: grapheme <- pop-grapheme-stack left
+  var left/eax: (addr code-point-utf8-stack) <- get self, left
+  var dummy/eax: code-point-utf8 <- pop-code-point-utf8-stack left
 }
 
-fn pop-after-gap _self: (addr gap-buffer) -> _/eax: grapheme {
+fn pop-after-gap _self: (addr gap-buffer) -> _/eax: code-point-utf8 {
   var self/eax: (addr gap-buffer) <- copy _self
-  var right/eax: (addr grapheme-stack) <- get self, right
-  var result/eax: grapheme <- pop-grapheme-stack right
+  var right/eax: (addr code-point-utf8-stack) <- get self, right
+  var result/eax: code-point-utf8 <- pop-code-point-utf8-stack right
   return result
 }
 
 fn gap-buffer-equal? _self: (addr gap-buffer), s: (addr array byte) -> _/eax: boolean {
   var self/esi: (addr gap-buffer) <- copy _self
-  # complication: graphemes may be multiple bytes
+  # complication: code-point-utf8s may be multiple bytes
   # so don't rely on length
   # instead turn the expected result into a stream and arrange to read from it in order
   var stream-storage: (stream byte 0x10/max-word-size)
   var expected-stream/ecx: (addr stream byte) <- address stream-storage
   write expected-stream, s
   # compare left
-  var left/edx: (addr grapheme-stack) <- get self, left
+  var left/edx: (addr code-point-utf8-stack) <- get self, left
   var result/eax: boolean <- prefix-match? left, expected-stream
   compare result, 0/false
   {
@@ -250,7 +250,7 @@ fn gap-buffer-equal? _self: (addr gap-buffer), s: (addr array byte) -> _/eax: bo
     return result
   }
   # compare right
-  var right/edx: (addr grapheme-stack) <- get self, right
+  var right/edx: (addr code-point-utf8-stack) <- get self, right
   result <- suffix-match? right, expected-stream
   compare result, 0/false
   {
@@ -267,10 +267,10 @@ fn test-gap-buffer-equal-from-end? {
   var g/esi: (addr gap-buffer) <- address _g
   initialize-gap-buffer g
   #
-  var c/eax: grapheme <- copy 0x61/a
-  add-grapheme-at-gap g, c
-  add-grapheme-at-gap g, c
-  add-grapheme-at-gap g, c
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  add-code-point-utf8-at-gap g, c
+  add-code-point-utf8-at-gap g, c
+  add-code-point-utf8-at-gap g, c
   # gap is at end (right is empty)
   var _result/eax: boolean <- gap-buffer-equal? g, "aaa"
   var result/eax: int <- copy _result
@@ -282,11 +282,11 @@ fn test-gap-buffer-equal-from-middle? {
   var g/esi: (addr gap-buffer) <- address _g
   initialize-gap-buffer g
   #
-  var c/eax: grapheme <- copy 0x61/a
-  add-grapheme-at-gap g, c
-  add-grapheme-at-gap g, c
-  add-grapheme-at-gap g, c
-  var dummy/eax: grapheme <- gap-left g
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  add-code-point-utf8-at-gap g, c
+  add-code-point-utf8-at-gap g, c
+  add-code-point-utf8-at-gap g, c
+  var dummy/eax: code-point-utf8 <- gap-left g
   # gap is in the middle
   var _result/eax: boolean <- gap-buffer-equal? g, "aaa"
   var result/eax: int <- copy _result
@@ -298,11 +298,11 @@ fn test-gap-buffer-equal-from-start? {
   var g/esi: (addr gap-buffer) <- address _g
   initialize-gap-buffer g
   #
-  var c/eax: grapheme <- copy 0x61/a
-  add-grapheme-at-gap g, c
-  add-grapheme-at-gap g, c
-  add-grapheme-at-gap g, c
-  var dummy/eax: grapheme <- gap-left g
+  var c/eax: code-point-utf8 <- copy 0x61/a
+  add-code-point-utf8-at-gap g, c
+  add-code-point-utf8-at-gap g, c
+  add-code-point-utf8-at-gap g, c
+  var dummy/eax: code-point-utf8 <- gap-left g
   dummy <- gap-left g
   dummy <- gap-left g
   # gap is at the start
@@ -319,25 +319,25 @@ fn copy-gap-buffer _src-ah: (addr handle gap-buffer), _dest-ah: (addr handle gap
   var dest-ah/eax: (addr handle gap-buffer) <- copy _dest-ah
   var _dest-a/eax: (addr gap-buffer) <- lookup *dest-ah
   var dest-a/edi: (addr gap-buffer) <- copy _dest-a
-  # copy left grapheme-stack
-  var src/ecx: (addr grapheme-stack) <- get src-a, left
-  var dest/edx: (addr grapheme-stack) <- get dest-a, left
-  copy-grapheme-stack src, dest
-  # copy right grapheme-stack
+  # copy left code-point-utf8-stack
+  var src/ecx: (addr code-point-utf8-stack) <- get src-a, left
+  var dest/edx: (addr code-point-utf8-stack) <- get dest-a, left
+  copy-code-point-utf8-stack src, dest
+  # copy right code-point-utf8-stack
   src <- get src-a, right
   dest <- get dest-a, right
-  copy-grapheme-stack src, dest
+  copy-code-point-utf8-stack src, dest
 }
 
 fn gap-buffer-is-decimal-integer? _self: (addr gap-buffer) -> _/eax: boolean {
   var self/esi: (addr gap-buffer) <- copy _self
-  var curr/ecx: (addr grapheme-stack) <- get self, left
-  var result/eax: boolean <- grapheme-stack-is-decimal-integer? curr
+  var curr/ecx: (addr code-point-utf8-stack) <- get self, left
+  var result/eax: boolean <- code-point-utf8-stack-is-decimal-integer? curr
   {
     compare result, 0/false
     break-if-=
     curr <- get self, right
-    result <- grapheme-stack-is-decimal-integer? curr
+    result <- code-point-utf8-stack-is-decimal-integer? curr
   }
   return result
 }
diff --git a/linux/tile/grapheme-stack.mu b/linux/tile/grapheme-stack.mu
index 0ea59ae5..c7565a1c 100644
--- a/linux/tile/grapheme-stack.mu
+++ b/linux/tile/grapheme-stack.mu
@@ -1,24 +1,24 @@
-type grapheme-stack {
-  data: (handle array grapheme)
+type code-point-utf8-stack {
+  data: (handle array code-point-utf8)
   top: int
 }
 
-fn initialize-grapheme-stack _self: (addr grapheme-stack), n: int {
-  var self/esi: (addr grapheme-stack) <- copy _self
-  var d/edi: (addr handle array grapheme) <- get self, data
+fn initialize-code-point-utf8-stack _self: (addr code-point-utf8-stack), n: int {
+  var self/esi: (addr code-point-utf8-stack) <- copy _self
+  var d/edi: (addr handle array code-point-utf8) <- get self, data
   populate d, n
   var top/eax: (addr int) <- get self, top
   copy-to *top, 0
 }
 
-fn clear-grapheme-stack _self: (addr grapheme-stack) {
-  var self/esi: (addr grapheme-stack) <- copy _self
+fn clear-code-point-utf8-stack _self: (addr code-point-utf8-stack) {
+  var self/esi: (addr code-point-utf8-stack) <- copy _self
   var top/eax: (addr int) <- get self, top
   copy-to *top, 0
 }
 
-fn grapheme-stack-empty? _self: (addr grapheme-stack) -> _/eax: boolean {
-  var self/esi: (addr grapheme-stack) <- copy _self
+fn code-point-utf8-stack-empty? _self: (addr code-point-utf8-stack) -> _/eax: boolean {
+  var self/esi: (addr code-point-utf8-stack) <- copy _self
   var top/eax: (addr int) <- get self, top
   compare *top, 0
   {
@@ -28,20 +28,20 @@ fn grapheme-stack-empty? _self: (addr grapheme-stack) -> _/eax: boolean {
   return 0/false
 }
 
-fn push-grapheme-stack _self: (addr grapheme-stack), _val: grapheme {
-  var self/esi: (addr grapheme-stack) <- copy _self
+fn push-code-point-utf8-stack _self: (addr code-point-utf8-stack), _val: code-point-utf8 {
+  var self/esi: (addr code-point-utf8-stack) <- copy _self
   var top-addr/ecx: (addr int) <- get self, top
-  var data-ah/edx: (addr handle array grapheme) <- get self, data
-  var data/eax: (addr array grapheme) <- lookup *data-ah
+  var data-ah/edx: (addr handle array code-point-utf8) <- get self, data
+  var data/eax: (addr array code-point-utf8) <- lookup *data-ah
   var top/edx: int <- copy *top-addr
-  var dest-addr/edx: (addr grapheme) <- index data, top
-  var val/eax: grapheme <- copy _val
+  var dest-addr/edx: (addr code-point-utf8) <- index data, top
+  var val/eax: code-point-utf8 <- copy _val
   copy-to *dest-addr, val
   add-to *top-addr, 1
 }
 
-fn pop-grapheme-stack _self: (addr grapheme-stack) -> _/eax: grapheme {
-  var self/esi: (addr grapheme-stack) <- copy _self
+fn pop-code-point-utf8-stack _self: (addr code-point-utf8-stack) -> _/eax: code-point-utf8 {
+  var self/esi: (addr code-point-utf8-stack) <- copy _self
   var top-addr/ecx: (addr int) <- get self, top
   {
     compare *top-addr, 0
@@ -49,25 +49,25 @@ fn pop-grapheme-stack _self: (addr grapheme-stack) -> _/eax: grapheme {
     return -1
   }
   subtract-from *top-addr, 1
-  var data-ah/edx: (addr handle array grapheme) <- get self, data
-  var data/eax: (addr array grapheme) <- lookup *data-ah
+  var data-ah/edx: (addr handle array code-point-utf8) <- get self, data
+  var data/eax: (addr array code-point-utf8) <- lookup *data-ah
   var top/edx: int <- copy *top-addr
-  var result-addr/eax: (addr grapheme) <- index data, top
+  var result-addr/eax: (addr code-point-utf8) <- index data, top
   return *result-addr
 }
 
-fn copy-grapheme-stack _src: (addr grapheme-stack), dest: (addr grapheme-stack) {
-  var src/esi: (addr grapheme-stack) <- copy _src
-  var data-ah/edi: (addr handle array grapheme) <- get src, data
-  var _data/eax: (addr array grapheme) <- lookup *data-ah
-  var data/edi: (addr array grapheme) <- copy _data
+fn copy-code-point-utf8-stack _src: (addr code-point-utf8-stack), dest: (addr code-point-utf8-stack) {
+  var src/esi: (addr code-point-utf8-stack) <- copy _src
+  var data-ah/edi: (addr handle array code-point-utf8) <- get src, data
+  var _data/eax: (addr array code-point-utf8) <- lookup *data-ah
+  var data/edi: (addr array code-point-utf8) <- copy _data
   var top-addr/ecx: (addr int) <- get src, top
   var i/eax: int <- copy 0
   {
     compare i, *top-addr
     break-if->=
-    var g/edx: (addr grapheme) <- index data, i
-    push-grapheme-stack dest, *g
+    var g/edx: (addr code-point-utf8) <- index data, i
+    push-code-point-utf8-stack dest, *g
     i <- increment
     loop
   }
@@ -75,18 +75,18 @@ fn copy-grapheme-stack _src: (addr grapheme-stack), dest: (addr grapheme-stack)
 
 # dump stack to screen from bottom to top
 # don't move the cursor or anything
-fn render-stack-from-bottom _self: (addr grapheme-stack), screen: (addr screen) {
-  var self/esi: (addr grapheme-stack) <- copy _self
-  var data-ah/edi: (addr handle array grapheme) <- get self, data
-  var _data/eax: (addr array grapheme) <- lookup *data-ah
-  var data/edi: (addr array grapheme) <- copy _data
+fn render-stack-from-bottom _self: (addr code-point-utf8-stack), screen: (addr screen) {
+  var self/esi: (addr code-point-utf8-stack) <- copy _self
+  var data-ah/edi: (addr handle array code-point-utf8) <- get self, data
+  var _data/eax: (addr array code-point-utf8) <- lookup *data-ah
+  var data/edi: (addr array code-point-utf8) <- copy _data
   var top-addr/ecx: (addr int) <- get self, top
   var i/eax: int <- copy 0
   {
     compare i, *top-addr
     break-if->=
-    var g/edx: (addr grapheme) <- index data, i
-    print-grapheme screen, *g
+    var g/edx: (addr code-point-utf8) <- index data, i
+    print-code-point-utf8 screen, *g
     i <- increment
     loop
   }
@@ -94,19 +94,19 @@ fn render-stack-from-bottom _self: (addr grapheme-stack), screen: (addr screen)
 
 # dump stack to screen from top to bottom
 # don't move the cursor or anything
-fn render-stack-from-top _self: (addr grapheme-stack), screen: (addr screen) {
-  var self/esi: (addr grapheme-stack) <- copy _self
-  var data-ah/edi: (addr handle array grapheme) <- get self, data
-  var _data/eax: (addr array grapheme) <- lookup *data-ah
-  var data/edi: (addr array grapheme) <- copy _data
+fn render-stack-from-top _self: (addr code-point-utf8-stack), screen: (addr screen) {
+  var self/esi: (addr code-point-utf8-stack) <- copy _self
+  var data-ah/edi: (addr handle array code-point-utf8) <- get self, data
+  var _data/eax: (addr array code-point-utf8) <- lookup *data-ah
+  var data/edi: (addr array code-point-utf8) <- copy _data
   var top-addr/ecx: (addr int) <- get self, top
   var i/eax: int <- copy *top-addr
   i <- decrement
   {
     compare i, 0
     break-if-<
-    var g/edx: (addr grapheme) <- index data, i
-    print-grapheme screen, *g
+    var g/edx: (addr code-point-utf8) <- index data, i
+    print-code-point-utf8 screen, *g
     i <- decrement
     loop
   }
@@ -114,11 +114,11 @@ fn render-stack-from-top _self: (addr grapheme-stack), screen: (addr screen) {
 
 # compare from bottom
 # beware: modifies 'stream', which must be disposed of after a false result
-fn prefix-match? _self: (addr grapheme-stack), s: (addr stream byte) -> _/eax: boolean {
-  var self/esi: (addr grapheme-stack) <- copy _self
-  var data-ah/edi: (addr handle array grapheme) <- get self, data
-  var _data/eax: (addr array grapheme) <- lookup *data-ah
-  var data/edi: (addr array grapheme) <- copy _data
+fn prefix-match? _self: (addr code-point-utf8-stack), s: (addr stream byte) -> _/eax: boolean {
+  var self/esi: (addr code-point-utf8-stack) <- copy _self
+  var data-ah/edi: (addr handle array code-point-utf8) <- get self, data
+  var _data/eax: (addr array code-point-utf8) <- lookup *data-ah
+  var data/edi: (addr array code-point-utf8) <- copy _data
   var top-addr/ecx: (addr int) <- get self, top
   var i/ebx: int <- copy 0
   {
@@ -126,8 +126,8 @@ fn prefix-match? _self: (addr grapheme-stack), s: (addr stream byte) -> _/eax: b
     break-if->=
     # if curr != expected, return false
     {
-      var curr-a/edx: (addr grapheme) <- index data, i
-      var expected/eax: grapheme <- read-grapheme s
+      var curr-a/edx: (addr code-point-utf8) <- index data, i
+      var expected/eax: code-point-utf8 <- read-code-point-utf8 s
       {
         compare expected, *curr-a
         break-if-=
@@ -142,11 +142,11 @@ fn prefix-match? _self: (addr grapheme-stack), s: (addr stream byte) -> _/eax: b
 
 # compare from bottom
 # beware: modifies 'stream', which must be disposed of after a false result
-fn suffix-match? _self: (addr grapheme-stack), s: (addr stream byte) -> _/eax: boolean {
-  var self/esi: (addr grapheme-stack) <- copy _self
-  var data-ah/edi: (addr handle array grapheme) <- get self, data
-  var _data/eax: (addr array grapheme) <- lookup *data-ah
-  var data/edi: (addr array grapheme) <- copy _data
+fn suffix-match? _self: (addr code-point-utf8-stack), s: (addr stream byte) -> _/eax: boolean {
+  var self/esi: (addr code-point-utf8-stack) <- copy _self
+  var data-ah/edi: (addr handle array code-point-utf8) <- get self, data
+  var _data/eax: (addr array code-point-utf8) <- lookup *data-ah
+  var data/edi: (addr array code-point-utf8) <- copy _data
   var top-addr/eax: (addr int) <- get self, top
   var i/ebx: int <- copy *top-addr
   i <- decrement
@@ -154,8 +154,8 @@ fn suffix-match? _self: (addr grapheme-stack), s: (addr stream byte) -> _/eax: b
     compare i, 0
     break-if-<
     {
-      var curr-a/edx: (addr grapheme) <- index data, i
-      var expected/eax: grapheme <- read-grapheme s
+      var curr-a/edx: (addr code-point-utf8) <- index data, i
+      var expected/eax: code-point-utf8 <- read-code-point-utf8 s
       # if curr != expected, return false
       {
         compare expected, *curr-a
@@ -169,18 +169,18 @@ fn suffix-match? _self: (addr grapheme-stack), s: (addr stream byte) -> _/eax: b
   return 1   # true
 }
 
-fn grapheme-stack-is-decimal-integer? _self: (addr grapheme-stack) -> _/eax: boolean {
-  var self/esi: (addr grapheme-stack) <- copy _self
-  var data-ah/eax: (addr handle array grapheme) <- get self, data
-  var _data/eax: (addr array grapheme) <- lookup *data-ah
-  var data/edx: (addr array grapheme) <- copy _data
+fn code-point-utf8-stack-is-decimal-integer? _self: (addr code-point-utf8-stack) -> _/eax: boolean {
+  var self/esi: (addr code-point-utf8-stack) <- copy _self
+  var data-ah/eax: (addr handle array code-point-utf8) <- get self, data
+  var _data/eax: (addr array code-point-utf8) <- lookup *data-ah
+  var data/edx: (addr array code-point-utf8) <- copy _data
   var top-addr/ecx: (addr int) <- get self, top
   var i/ebx: int <- copy 0
   var result/eax: boolean <- copy 1/true
-  $grapheme-stack-is-integer?:loop: {
+  $code-point-utf8-stack-is-integer?:loop: {
     compare i, *top-addr
     break-if->=
-    var g/edx: (addr grapheme) <- index data, i
+    var g/edx: (addr code-point-utf8) <- index data, i
     result <- decimal-digit? *g
     compare result, 0/false
     break-if-=
diff --git a/linux/tile/main.mu b/linux/tile/main.mu
index e0daaf1b..f2561a28 100644
--- a/linux/tile/main.mu
+++ b/linux/tile/main.mu
@@ -54,7 +54,7 @@ fn interactive {
   initialize-environment env
   {
     render env
-    var key/eax: grapheme <- read-key-from-real-keyboard
+    var key/eax: code-point-utf8 <- read-key-from-real-keyboard
     compare key, 0x11/ctrl-q
     break-if-=
     process env, key
@@ -79,7 +79,7 @@ fn process-all env: (addr environment), cmds: (addr array byte) {
     var done?/eax: boolean <- stream-empty? cmds-stream-a
     compare done?, 0/false
     break-if-!=
-    var g/eax: grapheme <- read-grapheme cmds-stream-a
+    var g/eax: code-point-utf8 <- read-code-point-utf8 cmds-stream-a
     process env, g
     loop
   }
@@ -105,7 +105,7 @@ fn repl {
       var done?/eax: boolean <- stream-empty? line
       compare done?, 0/false
       break-if-!=
-      var g/eax: grapheme <- read-grapheme line
+      var g/eax: code-point-utf8 <- read-code-point-utf8 line
       process env, g
       loop
     }
diff --git a/linux/tile/surface.mu b/linux/tile/surface.mu
index 2e353022..e86e22b9 100644
--- a/linux/tile/surface.mu
+++ b/linux/tile/surface.mu
@@ -111,10 +111,10 @@ fn print-surface-cell-at _self: (addr surface), screen-row: int, screen-col: int
   compare idx, 0
   {
     break-if->=
-    var space/ecx: grapheme <- copy 0x20
+    var space/ecx: code-point-utf8 <- copy 0x20
     var screen-ah/edi: (addr handle screen) <- get self, screen
     var screen/eax: (addr screen) <- lookup *screen-ah
-    print-grapheme screen, space
+    print-code-point-utf8 screen, space
     return
   }
   # otherwise print the appropriate screen-cell
@@ -156,9 +156,9 @@ fn print-screen-cell screen: (addr screen), _cell: (addr screen-cell) {
     break-if-=
     start-blinking screen
   }
-  var g/eax: (addr grapheme) <- get cell, data
-  print-grapheme screen, *g
-#?   var g2/eax: grapheme <- copy *g
+  var g/eax: (addr code-point-utf8) <- get cell, data
+  print-code-point-utf8 screen, *g
+#?   var g2/eax: code-point-utf8 <- copy *g
 #?   var g3/eax: int <- copy g2
 #?   print-int32-hex-to-real-screen g3
 #?   print-string-to-real-screen "\n"
@@ -264,7 +264,7 @@ fn num-lines in: (addr array byte) -> _/ecx: int {
     var done?/eax: boolean <- stream-empty? s-addr
     compare done?, 0/false
     break-if-!=
-    var g/eax: grapheme <- read-grapheme s-addr
+    var g/eax: code-point-utf8 <- read-code-point-utf8 s-addr
     compare g, 0xa/newline
     loop-if-!=
     result <- increment
@@ -282,7 +282,7 @@ fn first-line-length in: (addr array byte) -> _/edx: int {
     var done?/eax: boolean <- stream-empty? s-addr
     compare done?, 0/false
     break-if-!=
-    var g/eax: grapheme <- read-grapheme s-addr
+    var g/eax: code-point-utf8 <- read-code-point-utf8 s-addr
     compare g, 0xa/newline
     break-if-=
     result <- increment
@@ -301,12 +301,12 @@ fn fill-in _out: (addr array screen-cell), in: (addr array byte) {
     var done?/eax: boolean <- stream-empty? s-addr
     compare done?, 0/false
     break-if-!=
-    var g/eax: grapheme <- read-grapheme s-addr
+    var g/eax: code-point-utf8 <- read-code-point-utf8 s-addr
     compare g, 0xa/newline
     loop-if-=
     var offset/edx: (offset screen-cell) <- compute-offset out, idx
     var dest/edx: (addr screen-cell) <- index out, offset
-    var dest2/edx: (addr grapheme) <- get dest, data
+    var dest2/edx: (addr code-point-utf8) <- get dest, data
     copy-to *dest2, g
     idx <- increment
     loop
diff --git a/linux/tile/value.mu b/linux/tile/value.mu
index 8bd01676..0eacd8be 100644
--- a/linux/tile/value.mu
+++ b/linux/tile/value.mu
@@ -90,15 +90,15 @@ fn render-number screen: (addr screen), val: float, top-level?: boolean {
     fg <- copy 0
   }
   start-color screen, fg, bg
-  print-grapheme screen, 0x20/space
+  print-code-point-utf8 screen, 0x20/space
   print-float-decimal-approximate screen, val, 3
-  print-grapheme screen, 0x20/space
+  print-code-point-utf8 screen, 0x20/space
 }
 
 fn render-array-at screen: (addr screen), row: int, col: int, _a: (addr array value) {
   start-color screen, 0xf2, 7
   # don't surround in spaces
-  print-grapheme screen, 0x5b/[
+  print-code-point-utf8 screen, 0x5b/[
   increment col
   var a/esi: (addr array value) <- copy _a
   var max/ecx: int <- length a
@@ -122,7 +122,7 @@ fn render-array-at screen: (addr screen), row: int, col: int, _a: (addr array va
     i <- increment
     loop
   }
-  print-grapheme screen, 0x5d/]
+  print-code-point-utf8 screen, 0x5d/]
 }
 
 fn render-screen screen: (addr screen), row: int, col: int, _target-screen: (addr screen) {
@@ -179,13 +179,13 @@ fn print-screen-cell-of-fake-screen screen: (addr screen), _target: (addr screen
     start-blinking screen
     start-color screen, 0, 1
   }
-  var g/eax: grapheme <- screen-grapheme-at target, row, col
+  var g/eax: code-point-utf8 <- screen-code-point-utf8-at target, row, col
   {
     compare g, 0
     break-if-!=
     g <- copy 0x20/space
   }
-  print-grapheme screen, g
+  print-code-point-utf8 screen, g
   reset-formatting screen
 }
 
diff --git a/linux/tile/word.mu b/linux/tile/word.mu
index b4f5000b..3b4851f0 100644
--- a/linux/tile/word.mu
+++ b/linux/tile/word.mu
@@ -58,15 +58,15 @@ fn move-word-contents _src-ah: (addr handle word), _dest-ah: (addr handle word)
   cursor-to-start src
   var src-data-ah/eax: (addr handle gap-buffer) <- get src, scalar-data
   var src-data/eax: (addr gap-buffer) <- lookup *src-data-ah
-  var src-stack/ecx: (addr grapheme-stack) <- get src-data, right
+  var src-stack/ecx: (addr code-point-utf8-stack) <- get src-data, right
   {
-    var done?/eax: boolean <- grapheme-stack-empty? src-stack
+    var done?/eax: boolean <- code-point-utf8-stack-empty? src-stack
     compare done?, 0/false
     break-if-!=
-    var g/eax: grapheme <- pop-grapheme-stack src-stack
-#?     print-grapheme 0, g
+    var g/eax: code-point-utf8 <- pop-code-point-utf8-stack src-stack
+#?     print-code-point-utf8 0, g
 #?     print-string 0, "\n"
-    add-grapheme-to-word dest, g
+    add-code-point-utf8-to-word dest, g
     loop
   }
 }
@@ -79,17 +79,17 @@ fn copy-word-contents-before-cursor _src-ah: (addr handle word), _dest-ah: (addr
   var src/eax: (addr word) <- lookup *src-ah
   var src-data-ah/eax: (addr handle gap-buffer) <- get src, scalar-data
   var src-data/eax: (addr gap-buffer) <- lookup *src-data-ah
-  var src-stack/ecx: (addr grapheme-stack) <- get src-data, left
-  var src-stack-data-ah/eax: (addr handle array grapheme) <- get src-stack, data
-  var _src-stack-data/eax: (addr array grapheme) <- lookup *src-stack-data-ah
-  var src-stack-data/edx: (addr array grapheme) <- copy _src-stack-data
+  var src-stack/ecx: (addr code-point-utf8-stack) <- get src-data, left
+  var src-stack-data-ah/eax: (addr handle array code-point-utf8) <- get src-stack, data
+  var _src-stack-data/eax: (addr array code-point-utf8) <- lookup *src-stack-data-ah
+  var src-stack-data/edx: (addr array code-point-utf8) <- copy _src-stack-data
   var top-addr/ecx: (addr int) <- get src-stack, top
   var i/eax: int <- copy 0
   {
     compare i, *top-addr
     break-if->=
-    var g/edx: (addr grapheme) <- index src-stack-data, i
-    add-grapheme-to-word dest, *g
+    var g/edx: (addr code-point-utf8) <- index src-stack-data, i
+    add-code-point-utf8-to-word dest, *g
     i <- increment
     loop
   }
@@ -129,27 +129,27 @@ fn final-word _in: (addr handle word), out: (addr handle word) {
   copy-object curr-ah, out  # modify 'out' right at the end, just in case it's same as 'in'
 }
 
-fn first-grapheme _self: (addr word) -> _/eax: grapheme {
+fn first-code-point-utf8 _self: (addr word) -> _/eax: code-point-utf8 {
   var self/esi: (addr word) <- copy _self
   var data-ah/eax: (addr handle gap-buffer) <- get self, scalar-data
   var data/eax: (addr gap-buffer) <- lookup *data-ah
-  var result/eax: grapheme <- first-grapheme-in-gap-buffer data
+  var result/eax: code-point-utf8 <- first-code-point-utf8-in-gap-buffer data
   return result
 }
 
-fn grapheme-before-cursor _self: (addr word) -> _/eax: grapheme {
+fn code-point-utf8-before-cursor _self: (addr word) -> _/eax: code-point-utf8 {
   var self/esi: (addr word) <- copy _self
   var data-ah/eax: (addr handle gap-buffer) <- get self, scalar-data
   var data/eax: (addr gap-buffer) <- lookup *data-ah
-  var result/eax: grapheme <- grapheme-before-cursor-in-gap-buffer data
+  var result/eax: code-point-utf8 <- code-point-utf8-before-cursor-in-gap-buffer data
   return result
 }
 
-fn add-grapheme-to-word _self: (addr word), c: grapheme {
+fn add-code-point-utf8-to-word _self: (addr word), c: code-point-utf8 {
   var self/esi: (addr word) <- copy _self
   var data-ah/eax: (addr handle gap-buffer) <- get self, scalar-data
   var data/eax: (addr gap-buffer) <- lookup *data-ah
-  add-grapheme-at-gap data, c
+  add-code-point-utf8-at-gap data, c
 }
 
 fn cursor-at-start? _self: (addr word) -> _/eax: boolean {
@@ -172,14 +172,14 @@ fn cursor-left _self: (addr word) {
   var self/esi: (addr word) <- copy _self
   var data-ah/eax: (addr handle gap-buffer) <- get self, scalar-data
   var data/eax: (addr gap-buffer) <- lookup *data-ah
-  var dummy/eax: grapheme <- gap-left data
+  var dummy/eax: code-point-utf8 <- gap-left data
 }
 
 fn cursor-right _self: (addr word) {
   var self/esi: (addr word) <- copy _self
   var data-ah/eax: (addr handle gap-buffer) <- get self, scalar-data
   var data/eax: (addr gap-buffer) <- lookup *data-ah
-  var dummy/eax: grapheme <- gap-right data
+  var dummy/eax: code-point-utf8 <- gap-right data
 }
 
 fn cursor-to-start _self: (addr word) {
@@ -211,11 +211,11 @@ fn delete-before-cursor _self: (addr word) {
   delete-before-gap data
 }
 
-fn pop-after-cursor _self: (addr word) -> _/eax: grapheme {
+fn pop-after-cursor _self: (addr word) -> _/eax: code-point-utf8 {
   var self/esi: (addr word) <- copy _self
   var data-ah/eax: (addr handle gap-buffer) <- get self, scalar-data
   var data/eax: (addr gap-buffer) <- lookup *data-ah
-  var result/eax: grapheme <- pop-after-gap data
+  var result/eax: code-point-utf8 <- pop-after-gap data
   return result
 }
 
@@ -553,14 +553,14 @@ fn parse-words in: (addr array byte), out-ah: (addr handle word) {
     var done?/eax: boolean <- stream-empty? in-stream-a
     compare done?, 0/false
     break-if-!=
-    var _g/eax: grapheme <- read-grapheme in-stream-a
-    var g/ecx: grapheme <- copy _g
+    var _g/eax: code-point-utf8 <- read-code-point-utf8 in-stream-a
+    var g/ecx: code-point-utf8 <- copy _g
     # if not space, insert
     compare g, 0x20/space
     {
       break-if-=
       var cursor-word/eax: (addr word) <- lookup *cursor-word-ah
-      add-grapheme-to-word cursor-word, g
+      add-code-point-utf8-to-word cursor-word, g
       loop $parse-words:loop
     }
     # otherwise insert word after and move cursor to it
diff --git a/linux/vocabulary.md b/linux/vocabulary.md
index f1c6e3a5..2eefae33 100644
--- a/linux/vocabulary.md
+++ b/linux/vocabulary.md
@@ -206,8 +206,8 @@ doesn't yet parse floating-point literals:
 - `print-int32-buffered`: int -> buffered-file
   - textual representation in hex, including '0x' prefix
 
-- `write-grapheme`: grapheme -> stream
-- `to-grapheme`: code-point -> grapheme
+- `write-code-point-utf8`: code-point-utf8 -> stream
+- `to-utf8`: code-point -> code-point-utf8
 
 - `write-float-decimal-approximate`: float, precision: int -> stream
 
@@ -226,8 +226,8 @@ there isn't enough room in the destination stream.
 - `read-line-buffered`: buffered-file -> stream
   - Will abort the entire program if there isn't enough room.
 
-- `read-grapheme`: stream -> grapheme
-- `read-grapheme-buffered`: buffered-file -> grapheme
+- `read-code-point-utf8`: stream -> code-point-utf8
+- `read-code-point-utf8-buffered`: buffered-file -> code-point-utf8
 
 - `read-lines`: buffered-file -> array of strings
 
@@ -268,7 +268,7 @@ Unix terminal properties supported by almost all modern terminal emulators.
 
 - `print-string`: string -> screen
 - `print-stream`
-- `print-grapheme`
+- `print-code-point-utf8`
 - `print-code-point`
 - `print-int32-hex`
 - `print-int32-decimal`
@@ -290,7 +290,7 @@ manipulated.
 
 Assertions for tests:
 
-- `screen-grapheme-at`
+- `screen-code-point-utf8-at`
 - `screen-color-at`
 - `screen-background-color-at`
 - `screen-bold-at?`
-- 
cgit 1.4.1-2-gfad0