; Mu: An exploration on making the global structure of programs more accessible. ; ; "Is it a language, or an operating system, or a virtual machine? Mu." ; (with apologies to Robert Pirsig: http://en.wikipedia.org/wiki/Mu_%28negative%29#In_popular_culture) ; ;; Motivation ; ; I want to live in a world where I can have an itch to tweak a program, clone ; its open-source repository, orient myself on how it's organized, and make ; the simple change I envisioned, all in an afternoon. This codebase tries to ; make this possible for its readers. (More details: http://akkartik.name/about) ; ; What helps comprehend the global structure of programs? For starters, let's ; enumerate what doesn't: idiomatic code, adherence to a style guide or naming ; convention, consistent indentation, API documentation for each class, etc. ; These conventional considerations improve matters in the small, but don't ; help understand global organization. They help existing programmers manage ; day-to-day operations, but they can't turn outsider programmers into ; insiders. (Elaboration: http://akkartik.name/post/readable-bad) ; ; In my experience, two things have improved matters so far: version control ; and automated tests. Version control lets me rewind back to earlier, simpler ; times when the codebase was simpler, when its core skeleton was easier to ; ascertain. Indeed, arguably what came first is by definition the skeleton of ; a program, modulo major rewrites. Once you understand the skeleton, it ; becomes tractable to 'play back' later major features one by one. (Previous ; project that fleshed out this idea: http://akkartik.name/post/wart-layers) ; ; The second and biggest boost to comprehension comes from tests. Tests are ; good for writers for well-understood reasons: they avoid regressions, and ; they can influence code to be more decoupled and easier to change. In ; addition, tests are also good for the outsider reader because they permit ; active reading. If you can't build a program and run its tests it can't help ; you understand it. It hangs limp at best, and might even be actively ; misleading. If you can run its tests, however, it comes alive. You can step ; through scenarios in a debugger. You can add logging and scan logs to make ; sense of them. You can run what-if scenarios: "why is this line not written ; like this?" Make a change, rerun tests: "Oh, that's why." (Elaboration: ; http://akkartik.name/post/literate-programming) ; ; However, tests are only useful to the extent that they exist. Think back to ; your most recent codebase. Do you feel comfortable releasing a new version ; just because the tests pass? I'm not aware of any such project. There's just ; too many situations envisaged by the authors that were never encoded in a ; test. Even disciplined authors can't test for performance or race conditions ; or fault tolerance. If a line is phrased just so because of some subtle ; performance consideration, it's hard to communicate to newcomers. ; ; This isn't an arcane problem, and it isn't just a matter of altruism. As ; more and more such implicit considerations proliferate, and as the original ; authors are replaced by latecomers for day-to-day operations, knowledge is ; actively forgotten and lost. The once-pristine codebase turns into legacy ; code that is hard to modify without expensive and stress-inducing ; regressions. ; ; How to write tests for performance, fault tolerance, race conditions, etc.? ; How can we state and verify that a codepath doesn't ever perform memory ; allocation, or write to disk? It requires better, more observable primitives ; than we currently have. Modern operating systems have their roots in the ; 70s. Their interfaces were not designed to be testable. They provide no way ; to simulate a full disk, or a specific sequence of writes from different ; threads. We need something better. ; ; This project tries to move, groping, towards that 'something better', a ; platform that is both thoroughly tested and allows programs written for it ; to be thoroughly tested. It tries to answer the question: ; ; If Denis Ritchie and Ken Thompson were to set out today to co-design unix ; and C, knowing what we know about automated tests, what would they do ; differently? ; ; To try to impose *some* constraints on this gigantic yak-shave, we'll try to ; keep both language and OS as simple as possible, focused entirely on ; permitting more kinds of tests, on first *collecting* all the information ; about implicit considerations in some form so that readers and tools can ; have at least some hope of making sense of it. ; ; The initial language will be just assembly. We'll try to make it convenient ; to program in with some simple localized rewrite rules inspired by lisp ; macros and literate programming. Programmers will have to do their own ; memory management and register allocation, but we'll provide libraries to ; help with them. ; ; The initial OS will provide just memory management and concurrency ; primitives. No users or permissions (we don't live on mainframes anymore), ; no kernel- vs user-mode, no virtual memory or process abstraction, all ; threads sharing a single address space (use VMs for security and ; sandboxing). The only use case we care about is getting a test harness to ; run some code, feed it data through blocking channels, stop it and observe ; its internals. The code under test is expected to cooperate in such testing, ; by logging important events for the test harness to observe. (More info: ; http://akkartik.name/post/tracing-tests) ; ; The common thread here is elimination of abstractions, and it's not an ; accident. Abstractions help insiders manage the evolution of a codebase, but ; they actively hinder outsiders in understanding it from scratch. This ; matters, because the funnel to turn outsiders into insiders is critical to ; the long-term life of a codebase. Perhaps authors should raise their ; estimation of the costs of abstraction, and go against their instincts for ; introducing it. That's what I'll be trying to do: question every abstraction ; before I introduce it. We'll see how it goes. ; --- ;; Getting started ; ; Mu is currently built atop Racket and Arc, but this is temporary and ; contingent. We want to keep our options open, whether to port to a different ; host language, and easy to rewrite to native code for any platform. So we'll ; try to avoid 'cheating': relying on the host platform for advanced ; functionality. ; ; Other than that, we'll say no more about the code, and focus in the rest of ; this file on the scenarios the code cares about. (selective-load "mu.arc" section-level) (ero "running tests in mu.ar.c.t (takes ~30s)") ;? (quit) (set allow-raw-addresses*) (section 20 ; Our language is assembly-like in that functions consist of series of ; statements, and statements consist of an operation and its arguments (input ; and output). ; ; oarg1, oarg2, ... <- op arg1, arg2, ... ; ; Args must be atomic, like an integer or a memory address, they can't be ; expressions doing arithmetic or function calls. But we can have any number ; of them. ; ; Since we're building on lisp, our code samples won't look quite like the ; idealized syntax above. For now they will look like this: ; ; (function f [ ; (oarg1 oarg2 ... <- op arg1 arg2 ...) ; ... ; ... ; ]) ; ; Each arg/oarg can contain metadata separated by slashes and colons. In this ; first example below, the only metadata is types: 'integer' for a memory ; location containing an integer, and 'literal' for a value included directly ; in code. (Assembly languages traditionally call them 'immediate' operands.) ; In the future a simple tool will check that the types line up as expected in ; each op. A different tool might add types where they aren't provided. ; Instead of a monolithic compiler I want to build simple, lightweight tools ; that can be combined in various ways, say for using different typecheckers ; in different subsystems. ; ; In our tests we'll define such mu functions using a call to 'add-code', so ; look for it when reading the code examples. Everything outside 'add-code' is ; just test-harness details that can be skipped at first. (reset) ;? (set dump-trace*) (new-trace "literal") (add-code '((function main [ (1:integer <- copy 23:literal) ]))) ;? (set dump-trace*) (run 'main) ;? (prn memory*) (when (~is memory*.1 23) (prn "F - 'copy' writes its lone 'arg' after the instruction name to its lone 'oarg' or output arg before the arrow. After this test, the value 23 is stored in memory address 1.")) ;? (reset) ;? 2 ;? (quit) ;? 2 ; Our basic arithmetic ops can operate on memory locations or literals. ; (Ignore hardware details like registers for now.) (reset) (new-trace "add") (add-code '((function main [ (1:integer <- copy 1:literal) (2:integer <- copy 3:literal) (3:integer <- add 1:integer 2:integer) ]))) (run 'main) ;? (prn memory*) (when (~iso memory* (obj 1 1 2 3 3 4)) (prn "F - 'add' operates on two addresses")) ;? (reset) ;? 1 ;? (quit) ;? 1 (reset) (new-trace "add-literal") (add-code '((function main [ (1:integer <- add 2:literal 3:literal) ]))) (run 'main) (when (~is memory*.1 5) (prn "F - ops can take 'literal' operands (but not return them)")) (reset) (new-trace "sub-literal") (add-code '((function main [ (1:integer <- subtract 1:literal 3:literal) ]))) (run 'main) ;? (prn memory*) (when (~is memory*.1 -2) (prn "F - 'subtract'")) (reset) (new-trace "mul-literal") (add-code '((function main [ (1:integer <- multiply 2:literal 3:literal) ]))) (run 'main) ;? (prn memory*) (when (~is memory*.1 6) (prn "F - 'multiply'")) (reset) (new-trace "div-literal") (add-code '((function main [ (1:integer <- divide 8:literal 3:literal) ]))) (run 'main) ;? (prn memory*) (when (~is memory*.1 (/ real.8 3)) (prn "F - 'divide'")) (reset) (new-trace "idiv-literal") (add-code '((function main [ (1:integer 2:integer <- divide-with-remainder 23:literal 6:literal) ]))) (run 'main) ;? (prn memory*) (when (~iso memory* (obj 1 3 2 5)) (prn "F - 'divide-with-remainder' performs integer division")) (reset) (new-trace "dummy-oarg") ;? (set dump-trace*) (add-code '((function main [ (_ 2:integer <- divide-with-remainder 23:literal 6:literal) ]))) (run 'main) (when (~iso memory* (obj 2 5)) (prn "F - '_' oarg can ignore some results")) ;? (quit) ; Basic boolean operations: and, or, not ; There are easy ways to encode booleans in binary, but we'll skip past those ; details for now. (reset) (new-trace "and-literal") (add-code '((function main [ (1:boolean <- and t:literal nil:literal) ]))) ;? (set dump-trace*) (run 'main) ;? (prn memory*) (when (~is memory*.1 nil) (prn "F - logical 'and' for booleans")) ; Basic comparison operations (reset) (new-trace "lt-literal") (add-code '((function main [ (1:boolean <- less-than 4:literal 3:literal) ]))) (run 'main) ;? (prn memory*) (when (~is memory*.1 nil) (prn "F - 'less-than' inequality operator")) (reset) (new-trace "le-literal-false") (add-code '((function main [ (1:boolean <- lesser-or-equal 4:literal 3:literal) ]))) (run 'main) ;? (prn memory*) (when (~is memory*.1 nil) (prn "F - 'lesser-or-equal'")) (reset) (new-trace "le-literal-true") (add-code '((function main [ (1:boolean <- lesser-or-equal 4:literal 4:literal) ]))) (run 'main) ;? (prn memory*) (when (~is memory*.1 t) (prn "F - 'lesser-or-equal' returns true for equal operands")) (reset) (new-trace "le-literal-true-2") (add-code '((function main [ (1:boolean <- lesser-or-equal 4:literal 5:literal) ]))) (run 'main) ;? (prn memory*) (when (~is memory*.1 t) (prn "F - 'lesser-or-equal' - 2")) ; Control flow operations: jump, jump-if
# Testable primitives for writing text to screen.
# (Mu doesn't yet have testable primitives for graphics.)
#
# Unlike the top-level, this text mode has no scrolling.

# coordinates here don't match top-level
# Here we're consistent with graphics mode. Top-level is consistent with
# terminal emulators.
type screen {
  width: int
  height: int
  data: (handle array screen-cell)
  cursor-x: int
  cursor-y: int
}

type screen-cell {
  data: grapheme
  color: int
}

fn initialize-screen screen: (addr screen), width: int, height: int {
  var screen-addr/esi: (addr screen) <- copy screen
  var tmp/eax: int <- copy 0
  var dest/edi: (addr int) <- copy 0
  # screen->width = width
  dest <- get screen-addr, width
  tmp <- copy width
  copy-to *dest, tmp
  # screen->height = height
  dest <- get screen-addr, height
  tmp <- copy height
  copy-to *dest, tmp
  # screen->data = new screen-cell[width*height]
  {
    var data-addr/edi: (addr handle array screen-cell) <- get screen-addr, data
    tmp <- multiply width
    populate data-addr, tmp
  }
  # screen->cursor-x = 0
  dest <- get screen-addr, cursor-x
  copy-to *dest, 0
  # screen->cursor-y = 0
  dest <- get screen-addr, cursor-y
  copy-to *dest, 0
}

# in graphemes
fn screen-size screen: (addr screen) -> _/eax: int, _/ecx: int {
  var width/eax: int <- copy 0
  var height/ecx: int <- copy 0
  compare screen, 0
  {
    break-if-!=
    return 0x80, 0x30  # 128x48
  }
  # fake screen
  var screen-addr/esi: (addr screen) <- copy screen
  var tmp/edx: (addr int) <- get screen-addr, width
  width <- copy *tmp
  tmp <- get screen-addr, height
  height <- copy *tmp
  return width, height
}

# testable screen primitive
# background color isn't configurable yet
fn draw-grapheme screen: (addr screen), g: grapheme, x: int, y: int, color: int {
  {
    compare screen, 0
    break-if-!=
    draw-grapheme-on-real-screen g, x, y, color, 0
    return
  }
  # fake screen
  var screen-addr/esi: (addr screen) <- copy screen
  var idx/ecx: int <- screen-cell-index screen-addr, x, y
  var data-ah/eax: (addr handle array screen-cell) <- get screen-addr, data
  var data/eax: (addr array screen-cell) <- lookup *data-ah
  var offset/ecx: (offset screen-cell) <- compute-offset data, idx
  var dest-cell/ecx: (addr screen-cell) <- index data, offset
  var dest-grapheme/eax: (addr grapheme) <- get dest-cell, data
  var g2/edx: grapheme <- copy g
  copy-to *dest-grapheme, g2
  var dest-color/eax: (addr int) <- get dest-cell, color
  var color2/edx: grapheme <- copy color
  copy-to *dest-color, color2
}

fn screen-cell-index screen-on-stack: (addr screen), x: int, y: int -> _/ecx: int {
  var screen/esi: (addr screen) <- copy screen-on-stack
  var height-addr/eax: (addr int) <- get screen, height
  var result/ecx: int <- copy y
  result <- multiply *height-addr
  result <- add x
  return result
}

fn cursor-position screen: (addr screen) -> _/eax: int, _/ecx: int {
  {
    compare screen, 0
    break-if-!=
    var x/eax: int <- copy 0
    var y/ecx: int <- copy 0
    x, y <- cursor-position-on-real-screen
    return x, y
  }
  # fake screen
  var screen-addr/esi: (addr screen) <- copy screen
  var cursor-x-addr/eax: (addr int) <- get screen-addr, cursor-x
  var cursor-y-addr/ecx: (addr int) <- get screen-addr, cursor-y
  return *cursor-x-addr, *cursor-y-addr
}

fn set-cursor-position screen: (addr screen), x: int, y: int {
  {
    compare screen, 0
    break-if-!=
    set-cursor-position-on-real-screen x, y
    return
  }
  # fake screen
  var screen-addr/esi: (addr screen) <- copy screen
  # ignore x < 0
  {
    compare x, 0
    break-if->=
    return
  }
  # ignore x >= width
  {
    var width-addr/eax: (addr int) <- get screen-addr, width
    var width/eax: int <- copy *width-addr
    compare x, width
    break-if-<=
    return
  }
  # ignore y < 0
  {
    compare y, 0
    break-if->=
    return
  }
  # ignore y >= height
  {
    var height-addr/eax: (addr int) <- get screen-addr, height
    var height/eax: int <- copy *height-addr
    compare y, height
    break-if-<
    return
  }
  # screen->cursor-x = x
  var dest/edi: (addr int) <- get screen-addr, cursor-x
  var src/eax: int <- copy x
  copy-to *dest, src
  # screen->cursor-y = y
  dest <- get screen-addr, cursor-y
  src <- copy y
  copy-to *dest, src
}

fn show-cursor screen: (addr screen), g: grapheme {
  {
    compare screen, 0
    break-if-!=
    show-cursor-on-real-screen g
    return
  }
  # fake screen
  var cursor-x/eax: int <- copy 0
  var cursor-y/ecx: int <- copy 0
  cursor-x, cursor-y <- cursor-position screen
  draw-grapheme screen, g, cursor-x, cursor-y, 0  # cursor color not tracked for fake screen
}

fn clear-screen screen: (addr screen) {
  {
    compare screen, 0
    break-if-!=
    clear-real-screen
    return
  }
  # fake screen
  var space/edi: grapheme <- copy 0x20
  set-cursor-position screen, 0, 0
  var screen-addr/esi: (addr screen) <- copy screen
  var y/eax: int <- copy 1
  var height/ecx: (addr int) <- get screen-addr, height
  {
    compare y, *height
    break-if->
    var x/edx: int <- copy 1
    var width/ebx: (addr int) <- get screen-addr, width
    {
      compare x, *width
      break-if->
      draw-grapheme screen, space, x, y, 0  # color=black
      x <- increment
      loop
    }
    y <- increment
    loop
  }
  set-cursor-position screen, 0, 0
}

# there's no grapheme that guarantees to cover every pixel, so we'll bump down
# to pixels for a real screen
fn clear-real-screen {
  var y/eax: int <- copy 0
  {
    compare y, 0x300  # screen-height = 768
    break-if->=
    var x/edx: int <- copy 0
    {
      compare x, 0x400  # screen-width = 1024
      break-if->=
      pixel-on-real-screen x, y, 0  # black
      x <- increment
      loop
    }
    y <- increment
    loop
  }
}

fn screen-grapheme-at screen-on-stack: (addr screen), x: int, y: int -> _/eax: grapheme {
  var screen-addr/esi: (addr screen) <- copy screen-on-stack
  var idx/ecx: int <- screen-cell-index screen-addr, x, y
  var result/eax: grapheme <- screen-grapheme-at-idx screen-addr, idx
  return result
}

fn screen-grapheme-at-idx screen-on-stack: (addr screen), idx-on-stack: int -> _/eax: grapheme {
  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
  return *src
}

fn screen-color-at screen-on-stack: (addr screen), x: int, y: int -> _/eax: int {
  var screen-addr/esi: (addr screen) <- copy screen-on-stack
  var idx/ecx: int <- screen-cell-index screen-addr, x, y
  var result/eax: int <- screen-color-at-idx screen-addr, idx
  return result
}

fn screen-color-at-idx screen-on-stack: (addr screen), idx-on-stack: int -> _/eax: int {
  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 int) <- get cell, color
  var result/eax: int <- copy *src
  return result
}
tion 'f1) (prn "F - sleeping routines can time out")) ;? (quit) (reset) (new-trace "sleep") (add-code '((function f1 [ (sleep for-some-cycles:literal 1:literal) (1:integer <- copy 0:literal) (1:integer <- copy 0:literal) ]) (function f2 [ (2:integer <- copy 0:literal) (2:integer <- copy 0:literal) ]))) ;? (= dump-trace* (obj whitelist '("run" "schedule"))) (run 'f1 'f2) (check-trace-contents "scheduler handles sleeping routines" '(("run" "f1 0") ("run" "sleeping until 2") ("schedule" "pushing f1 to sleep queue") ("run" "f2 0") ("run" "f2 1") ("schedule" "waking up f1") ("run" "f1 1") ("run" "f1 2") )) (reset) (new-trace "sleep-long") (add-code '((function f1 [ (sleep for-some-cycles:literal 20:literal) (1:integer <- copy 0:literal) (1:integer <- copy 0:literal) ]) (function f2 [ (2:integer <- copy 0:literal) (2:integer <- copy 0:literal) ]))) ;? (= dump-trace* (obj whitelist '("run" "schedule"))) (run 'f1 'f2) (check-trace-contents "scheduler progresses sleeping routines when there are no routines left to run" '(("run" "f1 0") ("run" "sleeping until 21") ("schedule" "pushing f1 to sleep queue") ("run" "f2 0") ("run" "f2 1") ("schedule" "waking up f1") ("run" "f1 1") ("run" "f1 2") )) (reset) (new-trace "sleep-location") (add-code '((function f1 [ ; waits for memory location 1 to be set, before computing its successor (1:integer <- copy 0:literal) (sleep until-location-changes:literal 1:integer) (2:integer <- add 1:integer 1:literal) ]) (function f2 [ (sleep for-some-cycles:literal 30:literal) (1:integer <- copy 3:literal) ; set to value ]))) ;? (= dump-trace* (obj whitelist '("run" "schedule"))) ;? (set dump-trace*) (run 'f1 'f2) ;? (prn int-canon.memory*) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (when (~is memory*.2 4) ; successor of value (prn "F - sleep can block on a memory location")) ;? (quit) (reset) (new-trace "sleep-scoped-location") (add-code '((function f1 [ ; waits for memory location 1 to be changed, before computing its successor (10:integer <- copy 5:literal) ; array of locals (default-space:space-address <- copy 10:literal) (1:integer <- copy 23:literal) ; really location 12 (sleep until-location-changes:literal 1:integer) (2:integer <- add 1:integer 1:literal) ]) (function f2 [ (sleep for-some-cycles:literal 30:literal) (12:integer <- copy 3:literal) ; set to value ]))) ;? (= dump-trace* (obj whitelist '("run" "schedule"))) (run 'f1 'f2) (when (~is memory*.13 4) ; successor of value (prn "F - sleep can block on a scoped memory location")) ;? (quit) (reset) (new-trace "fork") (add-code '((function f1 [ (1:integer <- copy 4:literal) ]) (function main [ (fork f1:fn) ]))) (run 'main) (when (~iso memory*.1 4) (prn "F - fork works")) (reset) (new-trace "fork-returns-id") (add-code '((function f1 [ (1:integer <- copy 4:literal) ]) (function main [ (2:integer <- fork f1:fn) ]))) (run 'main) ;? (prn memory*) (when (no memory*.2) (prn "F - fork returns a pid for the new routine")) (reset) (new-trace "fork-returns-unique-id") (add-code '((function f1 [ (1:integer <- copy 4:literal) ]) (function main [ (2:integer <- fork f1:fn) (3:integer <- fork f1:fn) ]))) (run 'main) (when (or (no memory*.2) (no memory*.3) (is memory*.2 memory*.3)) (prn "F - fork returns a unique pid everytime")) (reset) (new-trace "fork-with-args") (add-code '((function f1 [ (2:integer <- next-input) ]) (function main [ (fork f1:fn nil:literal/globals nil:literal/limit 4:literal) ]))) (run 'main) (when (~iso memory*.2 4) (prn "F - fork can pass args")) (reset) (new-trace "fork-copies-args") (add-code '((function f1 [ (2:integer <- next-input) ]) (function main [ (default-space:space-address <- new space:literal 5:literal) (x:integer <- copy 4:literal) (fork f1:fn nil:literal/globals nil:literal/limit x:integer) (x:integer <- copy 0:literal) ; should be ignored ]))) (run 'main) (when (~iso memory*.2 4) (prn "F - fork passes args by value")) (reset) (new-trace "fork-global") (add-code '((function f1 [ (1:integer/raw <- copy 2:integer/space:global) ]) (function main [ (default-space:space-address <- new space:literal 5:literal) (2:integer <- copy 4:literal) (fork f1:fn default-space:space-address/globals nil:literal/limit) ]))) (run 'main) (each routine completed-routines* (awhen rep.routine!error (prn "error - " it))) (when (~iso memory*.1 4) (prn "F - fork can take a space of global variables to access")) (reset) (new-trace "fork-limit") (add-code '((function f1 [ { begin (loop) } ]) (function main [ (fork f1:fn nil:literal/globals 30:literal/limit) ]))) (= scheduling-interval* 5) (run 'main) (each routine completed-routines* (awhen rep.routine!error (prn "error - " it))) (when (ran-to-completion 'f1) (prn "F - fork can specify a maximum cycle limit")) (reset) (new-trace "fork-then-wait") (add-code '((function f1 [ { begin (loop) } ]) (function main [ (1:integer/routine-id <- fork f1:fn nil:literal/globals 30:literal/limit) (sleep until-routine-done:literal 1:integer/routine-id) (2:integer <- copy 34:literal) ]))) (= scheduling-interval* 5) ;? (= dump-trace* (obj whitelist '("schedule"))) (run 'main) (each routine completed-routines* (awhen rep.routine!error (prn "error - " it))) (check-trace-contents "scheduler orders functions correctly" '(("schedule" "pushing main to sleep queue") ("schedule" "scheduling f1") ("schedule" "ran out of time") ("schedule" "waking up main") )) ;? (quit) ; todo: Haven't yet written several tests ; that restarting a routine works ; when it died ; when it timed out ; when it completed ; running multiple routines in tandem ; first example using these features: read-move-incomplete in chessboard-cursor.arc.t ; The scheduler needs to keep track of the call stack for each routine. ; Eventually we'll want to save this information in mu's address space itself, ; along with the types array, the magic buffers for args and oargs, and so on. ; ; Eventually we want the right stack-management primitives to build delimited ; continuations in mu. ; Routines can throw errors. (reset) (new-trace "array-bounds-check") (add-code '((function main [ (1:integer <- copy 2:literal) (2:integer <- copy 23:literal) (3:integer <- copy 24:literal) (4:integer <- index 1:integer-array 2:literal) ]))) ;? (set dump-trace*) (run 'main) ;? (prn memory*) (let routine (car completed-routines*) (when (no rep.routine!error) (prn "F - 'index' throws an error if out of bounds"))) ) ; section 20 (section 100 ;; Synchronization ; ; Mu synchronizes using channels rather than locks, like Erlang and Go. ; ; The two ends of a channel will usually belong to different routines, but ; each end should only be used by a single one. Don't try to read from or ; write to it from multiple routines at once. ; ; To avoid locking, writer and reader will never write to the same location. ; So channels will include fields in pairs, one for the writer and one for the ; reader. ; The core circular buffer contains values at index 'first-full' up to (but ; not including) index 'first-empty'. The reader always modifies it at ; first-full, while the writer always modifies it at first-empty. (reset) (new-trace "channel-new") (add-code '((function main [ (1:channel-address <- init-channel 3:literal) (2:integer <- get 1:channel-address/deref first-full:offset) (3:integer <- get 1:channel-address/deref first-free:offset) ]))) ;? (set dump-trace*) (run 'main) ;? (prn memory*) (when (or (~is 0 memory*.2) (~is 0 memory*.3)) (prn "F - 'init-channel' initializes 'first-full and 'first-free to 0")) (reset) (new-trace "channel-write") (add-code '((function main [ (1:channel-address <- init-channel 3:literal) (2:integer <- copy 34:literal) (3:tagged-value <- save-type 2:integer) (1:channel-address/deref <- write 1:channel-address 3:tagged-value) (5:integer <- get 1:channel-address/deref first-full:offset) (6:integer <- get 1:channel-address/deref first-free:offset) ]))) ;? (prn function*!write) ;? (set dump-trace*) ;? (= dump-trace* (obj blacklist '("sz" "mem" "addr" "array-len" "cvt0" "cvt1"))) ;? (= dump-trace* (obj whitelist '("jump"))) ;? (= dump-trace* (obj whitelist '("run" "reply"))) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) ;? (prn canon.memory*) (when (or (~is 0 memory*.5) (~is 1 memory*.6)) (prn "F - 'write' enqueues item to channel")) ;? (quit) (reset) (new-trace "channel-read") (add-code '((function main [ (1:channel-address <- init-channel 3:literal) (2:integer <- copy 34:literal) (3:tagged-value <- save-type 2:integer) (1:channel-address/deref <- write 1:channel-address 3:tagged-value) (5:tagged-value 1:channel-address/deref <- read 1:channel-address) (7:integer <- maybe-coerce 5:tagged-value integer:literal) (8:integer <- get 1:channel-address/deref first-full:offset) (9:integer <- get 1:channel-address/deref first-free:offset) ]))) ;? (set dump-trace*) ;? (= dump-trace* (obj blacklist '("sz" "mem" "addr" "array-len" "cvt0" "cvt1"))) (run 'main) ;? (prn int-canon.memory*) (when (~is memory*.7 34) (prn "F - 'read' returns written value")) (when (or (~is 1 memory*.8) (~is 1 memory*.9)) (prn "F - 'read' dequeues item from channel")) (reset) (new-trace "channel-write-wrap") (add-code '((function main [ ; channel with 1 slot (1:channel-address <- init-channel 1:literal) ; write a value (2:integer <- copy 34:literal) (3:tagged-value <- save-type 2:integer) (1:channel-address/deref <- write 1:channel-address 3:tagged-value) ; first-free will now be 1 (5:integer <- get 1:channel-address/deref first-free:offset) ; read one value (_ 1:channel-address/deref <- read 1:channel-address) ; write a second value; verify that first-free wraps around to 0. (1:channel-address/deref <- write 1:channel-address 3:tagged-value) (6:integer <- get 1:channel-address/deref first-free:offset) ]))) ;? (set dump-trace*) ;? (= dump-trace* (obj blacklist '("sz" "mem" "addr" "array-len" "cvt0" "cvt1"))) (run 'main) ;? (prn canon.memory*) (when (or (~is 1 memory*.5) (~is 0 memory*.6)) (prn "F - 'write' can wrap pointer back to start")) (reset) (new-trace "channel-read-wrap") (add-code '((function main [ ; channel with 1 slot (1:channel-address <- init-channel 1:literal) ; write a value (2:integer <- copy 34:literal) (3:tagged-value <- save-type 2:integer) (1:channel-address/deref <- write 1:channel-address 3:tagged-value) ; read one value (_ 1:channel-address/deref <- read 1:channel-address) ; first-full will now be 1 (5:integer <- get 1:channel-address/deref first-full:offset) ; write a second value (1:channel-address/deref <- write 1:channel-address 3:tagged-value) ; read second value; verify that first-full wraps around to 0. (_ 1:channel-address/deref <- read 1:channel-address) (6:integer <- get 1:channel-address/deref first-full:offset) ]))) ;? (set dump-trace*) ;? (= dump-trace* (obj blacklist '("sz" "mem" "addr" "array-len" "cvt0" "cvt1"))) (run 'main) ;? (prn canon.memory*) (when (or (~is 1 memory*.5) (~is 0 memory*.6)) (prn "F - 'read' can wrap pointer back to start")) (reset) (new-trace "channel-new-empty-not-full") (add-code '((function main [ (1:channel-address <- init-channel 3:literal) (2:boolean <- empty? 1:channel-address/deref) (3:boolean <- full? 1:channel-address/deref) ]))) ;? (set dump-trace*) (run 'main) ;? (prn memory*) (when (or (~is t memory*.2) (~is nil memory*.3)) (prn "F - a new channel is always empty, never full")) (reset) (new-trace "channel-write-not-empty") (add-code '((function main [ (1:channel-address <- init-channel 3:literal) (2:integer <- copy 34:literal) (3:tagged-value <- save-type 2:integer) (1:channel-address/deref <- write 1:channel-address 3:tagged-value) (5:boolean <- empty? 1:channel-address/deref) (6:boolean <- full? 1:channel-address/deref) ]))) ;? (set dump-trace*) (run 'main) ;? (prn memory*) (when (or (~is nil memory*.5) (~is nil memory*.6)) (prn "F - a channel after writing is never empty")) (reset) (new-trace "channel-write-full") (add-code '((function main [ (1:channel-address <- init-channel 1:literal) (2:integer <- copy 34:literal) (3:tagged-value <- save-type 2:integer) (1:channel-address/deref <- write 1:channel-address 3:tagged-value) (5:boolean <- empty? 1:channel-address/deref) (6:boolean <- full? 1:channel-address/deref) ]))) ;? (set dump-trace*) (run 'main) ;? (prn memory*) (when (or (~is nil memory*.5) (~is t memory*.6)) (prn "F - a channel after writing may be full")) (reset) (new-trace "channel-read-not-full") (add-code '((function main [ (1:channel-address <- init-channel 3:literal) (2:integer <- copy 34:literal) (3:tagged-value <- save-type 2:integer) (1:channel-address/deref <- write 1:channel-address 3:tagged-value) (1:channel-address/deref <- write 1:channel-address 3:tagged-value) (_ 1:channel-address/deref <- read 1:channel-address) (5:boolean <- empty? 1:channel-address/deref) (6:boolean <- full? 1:channel-address/deref) ]))) ;? (set dump-trace*) (run 'main) ;? (prn memory*) (when (or (~is nil memory*.5) (~is nil memory*.6)) (prn "F - a channel after reading is never full")) (reset) (new-trace "channel-read-empty") (add-code '((function main [ (1:channel-address <- init-channel 3:literal) (2:integer <- copy 34:literal) (3:tagged-value <- save-type 2:integer) (1:channel-address/deref <- write 1:channel-address 3:tagged-value) (_ 1:channel-address/deref <- read 1:channel-address) (5:boolean <- empty? 1:channel-address/deref) (6:boolean <- full? 1:channel-address/deref) ]))) ;? (set dump-trace*) (run 'main) ;? (prn memory*) (when (or (~is t memory*.5) (~is nil memory*.6)) (prn "F - a channel after reading may be empty")) ; The key property of channels; writing to a full channel blocks the current ; routine until it creates space. Ditto reading from an empty channel. (reset) (new-trace "channel-read-block") (add-code '((function main [ (1:channel-address <- init-channel 3:literal) ; channel is empty, but receives a read (2:tagged-value 1:channel-address/deref <- read 1:channel-address) ]))) ;? (set dump-trace*) ;? (= dump-trace* (obj whitelist '("run" "schedule"))) (run 'main) ;? (prn int-canon.memory*) ;? (prn sleeping-routines*) ;? (prn completed-routines*) ; read should cause the routine to sleep, and ; the sole sleeping routine should trigger the deadlock detector (let routine (car completed-routines*) (when (or (no routine) (no rep.routine!error) (~posmatch "deadlock" rep.routine!error)) (prn "F - 'read' on empty channel blocks (puts the routine to sleep until the channel gets data)"))) ;? (quit) (reset) (new-trace "channel-write-block") (add-code '((function main [ (1:channel-address <- init-channel 1:literal) (2:integer <- copy 34:literal) (3:tagged-value <- save-type 2:integer) (1:channel-address/deref <- write 1:channel-address 3:tagged-value) ; channel has capacity 1, but receives a second write (1:channel-address/deref <- write 1:channel-address 3:tagged-value) ]))) ;? (set dump-trace*) ;? (= dump-trace* (obj whitelist '("run" "schedule" "addr"))) (run 'main) ;? (prn int-canon.memory*) ;? (prn running-routines*) ;? (prn sleeping-routines*) ;? (prn completed-routines*) ; second write should cause the routine to sleep, and ; the sole sleeping routine should trigger the deadlock detector (let routine (car completed-routines*) (when (or (no routine) (no rep.routine!error) (~posmatch "deadlock" rep.routine!error)) (prn "F - 'write' on full channel blocks (puts the routine to sleep until the channel gets data)"))) ;? (quit) (reset) (new-trace "channel-handoff") (add-code '((function consumer [ (default-space:space-address <- new space:literal 30:literal) (chan:channel-address <- init-channel 3:literal) ; create a channel (fork producer:fn nil:literal/globals nil:literal/limit chan:channel-address) ; fork a routine to produce a value in it (1:tagged-value/raw <- read chan:channel-address) ; wait for input on channel ]) (function producer [ (default-space:space-address <- new space:literal 30:literal) (n:integer <- copy 24:literal) (ochan:channel-address <- next-input) (x:tagged-value <- save-type n:integer) (ochan:channel-address/deref <- write ochan:channel-address x:tagged-value) ]))) ;? (set dump-trace*) ;? (= dump-trace* (obj whitelist '("schedule" "run" "addr"))) ;? (= dump-trace* (obj whitelist '("-"))) (run 'consumer) ;? (prn memory*) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (when (~is 24 memory*.2) ; location 1 contains tagged-value x above (prn "F - channels are meant to be shared between routines")) ;? (quit) (reset) (new-trace "channel-handoff-routine") (add-code '((function consumer [ (default-space:space-address <- new space:literal 30:literal) (1:channel-address <- init-channel 3:literal) ; create a channel (fork producer:fn default-space:space-address/globals nil:literal/limit) ; pass it as a global to another routine (1:tagged-value/raw <- read 1:channel-address) ; wait for input on channel ]) (function producer [ (default-space:space-address <- new space:literal 30:literal) (n:integer <- copy 24:literal) (x:tagged-value <- save-type n:integer) (1:channel-address/space:global/deref <- write 1:channel-address/space:global x:tagged-value) ]))) (run 'consumer) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (when (~is 24 memory*.2) ; location 1 contains tagged-value x above (prn "F - channels are meant to be shared between routines")) ) ; section 100 (section 10 ;; Separating concerns ; ; Lightweight tools can also operate on quoted lists of statements surrounded ; by square brackets. In the example below, we mimic Go's 'defer' keyword ; using 'convert-quotes'. It lets us write code anywhere in a function, but ; have it run just before the function exits. Great for keeping code to ; reclaim memory or other resources close to the code to allocate it. (C++ ; programmers know this as RAII.) We'll use 'defer' when we build a memory ; deallocation routine like C's 'free'. ; ; More powerful reorderings are also possible like in Literate Programming or ; Aspect-Oriented Programming; one advantage of prohibiting arbitrarily nested ; code is that we can naturally name 'join points' wherever we want. (reset) (new-trace "convert-quotes-defer") (= traces* (queue)) (when (~iso (convert-quotes '((1:integer <- copy 4:literal) (defer [ (3:integer <- copy 6:literal) ]) (2:integer <- copy 5:literal))) '((1:integer <- copy 4:literal) (2:integer <- copy 5:literal) (3:integer <- copy 6:literal))) (prn "F - convert-quotes can handle 'defer'")) (reset) (new-trace "convert-quotes-defer-reply") (= traces* (queue)) (when (~iso (convert-quotes '((1:integer <- copy 0:literal) (defer [ (5:integer <- copy 0:literal) ]) (2:integer <- copy 0:literal) (reply) (3:integer <- copy 0:literal) (4:integer <- copy 0:literal))) '((1:integer <- copy 0:literal) (2:integer <- copy 0:literal) (5:integer <- copy 0:literal) (reply) (3:integer <- copy 0:literal) (4:integer <- copy 0:literal) (5:integer <- copy 0:literal))) (prn "F - convert-quotes inserts code at early exits")) (reset) (new-trace "convert-quotes-defer-reply-arg") (= traces* (queue)) (when (~iso (convert-quotes '((1:integer <- copy 0:literal) (defer [ (5:integer <- copy 0:literal) ]) (2:integer <- copy 0:literal) (reply 2:literal) (3:integer <- copy 0:literal) (4:integer <- copy 0:literal))) '((1:integer <- copy 0:literal) (2:integer <- copy 0:literal) (prepare-reply 2:literal) (5:integer <- copy 0:literal) (reply) (3:integer <- copy 0:literal) (4:integer <- copy 0:literal) (5:integer <- copy 0:literal))) (prn "F - convert-quotes inserts code at early exits")) (reset) (new-trace "convert-quotes-label") (= traces* (queue)) (when (~iso (convert-quotes '((1:integer <- copy 4:literal) foo (2:integer <- copy 5:literal))) '((1:integer <- copy 4:literal) foo (2:integer <- copy 5:literal))) (prn "F - convert-quotes can handle labels")) (reset) (new-trace "before") (= traces* (queue)) (add-code '((before label1 [ (2:integer <- copy 0:literal) ]))) (when (~iso (as cons before*!label1) '(; fragment ( (2:integer <- copy 0:literal)))) (prn "F - 'before' records fragments of code to insert before labels")) (when (~iso (insert-code '((1:integer <- copy 0:literal) label1 (3:integer <- copy 0:literal))) '((1:integer <- copy 0:literal) (2:integer <- copy 0:literal) label1 (3:integer <- copy 0:literal))) (prn "F - 'insert-code' can insert fragments before labels")) (reset) (new-trace "before-multiple") (= traces* (queue)) (add-code '((before label1 [ (2:integer <- copy 0:literal) ]) (before label1 [ (3:integer <- copy 0:literal) ]))) (when (~iso (as cons before*!label1) '(; fragment ( (2:integer <- copy 0:literal)) ( (3:integer <- copy 0:literal)))) (prn "F - 'before' records fragments in order")) (when (~iso (insert-code '((1:integer <- copy 0:literal) label1 (4:integer <- copy 0:literal))) '((1:integer <- copy 0:literal) (2:integer <- copy 0:literal) (3:integer <- copy 0:literal) label1 (4:integer <- copy 0:literal))) (prn "F - 'insert-code' can insert multiple fragments in order before label")) (reset) (new-trace "before-scoped") (= traces* (queue)) (add-code '((before f/label1 [ ; label1 only inside function f (2:integer <- copy 0:literal) ]))) (when (~iso (insert-code '((1:integer <- copy 0:literal) label1 (3:integer <- copy 0:literal)) 'f) '((1:integer <- copy 0:literal) (2:integer <- copy 0:literal) label1 (3:integer <- copy 0:literal))) (prn "F - 'insert-code' can insert fragments before labels just in specified functions")) (reset) (new-trace "before-scoped2") (= traces* (queue)) (add-code '((before f/label1 [ ; label1 only inside function f (2:integer <- copy 0:literal) ]))) (when (~iso (insert-code '((1:integer <- copy 0:literal) label1 (3:integer <- copy 0:literal))) '((1:integer <- copy 0:literal) label1 (3:integer <- copy 0:literal))) (prn "F - 'insert-code' ignores labels not in specified functions")) (reset) (new-trace "after") (= traces* (queue)) (add-code '((after label1 [ (2:integer <- copy 0:literal) ]))) (when (~iso (as cons after*!label1) '(; fragment ( (2:integer <- copy 0:literal)))) (prn "F - 'after' records fragments of code to insert after labels")) (when (~iso (insert-code '((1:integer <- copy 0:literal) label1 (3:integer <- copy 0:literal))) '((1:integer <- copy 0:literal) label1 (2:integer <- copy 0:literal) (3:integer <- copy 0:literal))) (prn "F - 'insert-code' can insert fragments after labels")) (reset) (new-trace "after-multiple") (= traces* (queue)) (add-code '((after label1 [ (2:integer <- copy 0:literal) ]) (after label1 [ (3:integer <- copy 0:literal) ]))) (when (~iso (as cons after*!label1) '(; fragment ( (3:integer <- copy 0:literal)) ( (2:integer <- copy 0:literal)))) (prn "F - 'after' records fragments in *reverse* order")) (when (~iso (insert-code '((1:integer <- copy 0:literal) label1 (4:integer <- copy 0:literal))) '((1:integer <- copy 0:literal) label1 (3:integer <- copy 0:literal) (2:integer <- copy 0:literal) (4:integer <- copy 0:literal))) (prn "F - 'insert-code' can insert multiple fragments in order after label")) (reset) (new-trace "before-after") (= traces* (queue)) (add-code '((before label1 [ (2:integer <- copy 0:literal) ]) (after label1 [ (3:integer <- copy 0:literal) ]))) (when (and (~iso (as cons before*!label1) '(; fragment ( (2:integer <- copy 0:literal)))) (~iso (as cons after*!label1) '(; fragment ( (3:integer <- copy 0:literal))))) (prn "F - 'before' and 'after' fragments work together")) (when (~iso (insert-code '((1:integer <- copy 0:literal) label1 (4:integer <- copy 0:literal))) '((1:integer <- copy 0:literal) (2:integer <- copy 0:literal) label1 (3:integer <- copy 0:literal) (4:integer <- copy 0:literal))) (prn "F - 'insert-code' can insert multiple fragments around label")) (reset) (new-trace "before-after-multiple") (= traces* (queue)) (add-code '((before label1 [ (2:integer <- copy 0:literal) (3:integer <- copy 0:literal) ]) (after label1 [ (4:integer <- copy 0:literal) ]) (before label1 [ (5:integer <- copy 0:literal) ]) (after label1 [ (6:integer <- copy 0:literal) (7:integer <- copy 0:literal) ]))) (when (or (~iso (as cons before*!label1) '(; fragment ( (2:integer <- copy 0:literal) (3:integer <- copy 0:literal)) ( (5:integer <- copy 0:literal)))) (~iso (as cons after*!label1) '(; fragment ( (6:integer <- copy 0:literal) (7:integer <- copy 0:literal)) ( (4:integer <- copy 0:literal))))) (prn "F - multiple 'before' and 'after' fragments at once")) (when (~iso (insert-code '((1:integer <- copy 0:literal) label1 (8:integer <- copy 0:literal))) '((1:integer <- copy 0:literal) (2:integer <- copy 0:literal) (3:integer <- copy 0:literal) (5:integer <- copy 0:literal) label1 (6:integer <- copy 0:literal) (7:integer <- copy 0:literal) (4:integer <- copy 0:literal) (8:integer <- copy 0:literal))) (prn "F - 'insert-code' can insert multiple fragments around label - 2")) (reset) (new-trace "before-after-independent") (= traces* (queue)) (when (~iso (do (reset) (add-code '((before label1 [ (2:integer <- copy 0:literal) ]) (after label1 [ (3:integer <- copy 0:literal) ]) (before label1 [ (4:integer <- copy 0:literal) ]) (after label1 [ (5:integer <- copy 0:literal) ]))) (list before*!label1 after*!label1)) (do (reset) (add-code '((before label1 [ (2:integer <- copy 0:literal) ]) (before label1 [ (4:integer <- copy 0:literal) ]) (after label1 [ (3:integer <- copy 0:literal) ]) (after label1 [ (5:integer <- copy 0:literal) ]))) (list before*!label1 after*!label1))) (prn "F - order matters between 'before' and between 'after' fragments, but not *across* 'before' and 'after' fragments")) (reset) (new-trace "before-after-braces") (= traces* (queue)) (= function* (table)) (add-code '((after label1 [ (1:integer <- copy 0:literal) ]) (function f1 [ { begin label1 } ]))) ;? (= dump-trace* (obj whitelist '("cn0"))) (freeze function*) (when (~iso function*!f1 '(label1 (((1 integer)) <- ((copy)) ((0 literal))))) (prn "F - before/after works inside blocks")) (reset) (new-trace "before-after-any-order") (= traces* (queue)) (= function* (table)) (add-code '((function f1 [ { begin label1 } ]) (after label1 [ (1:integer <- copy 0:literal) ]))) (freeze function*) (when (~iso function*!f1 '(label1 (((1 integer)) <- ((copy)) ((0 literal))))) (prn "F - before/after can come after the function they need to modify")) ;? (quit) (reset) (new-trace "multiple-defs") (= traces* (queue)) (= function* (table)) (add-code '((function f1 [ (1:integer <- copy 0:literal) ]) (function f1 [ (2:integer <- copy 0:literal) ]))) (freeze function*) (when (~iso function*!f1 '((((2 integer)) <- ((copy)) ((0 literal))) (((1 integer)) <- ((copy)) ((0 literal))))) (prn "F - multiple 'def' of the same function add clauses")) (reset) (new-trace "def!") (= traces* (queue)) (= function* (table)) (add-code '((function f1 [ (1:integer <- copy 0:literal) ]) (function! f1 [ (2:integer <- copy 0:literal) ]))) (freeze function*) (when (~iso function*!f1 '((((2 integer)) <- ((copy)) ((0 literal))))) (prn "F - 'def!' clears all previous clauses")) ) ; section 10 ;; --- (section 100 ; String utilities (reset) (new-trace "string-new") (add-code '((function main [ (1:string-address <- new string:literal 5:literal) ]))) (let routine make-routine!main (enq routine running-routines*) (let before rep.routine!alloc (run) (when (~iso rep.routine!alloc (+ before 5 1)) (prn "F - 'new' allocates arrays of bytes for strings")))) ; Convenience: initialize strings using string literals (reset) (new-trace "string-literal") (add-code '((function main [ (1:string-address <- new "hello") ]))) (let routine make-routine!main (enq routine running-routines*) (let before rep.routine!alloc ;? (set dump-trace*) ;? (= dump-trace* (obj whitelist '("schedule" "run" "addr"))) (run) (when (~iso rep.routine!alloc (+ before 5 1)) (prn "F - 'new' allocates arrays of bytes for string literals")) (when (~memory-contains-array before "hello") (prn "F - 'new' initializes allocated memory to string literal")))) (reset) (new-trace "string-equal") (add-code '((function main [ (1:string-address <- new "hello") (2:string-address <- new "hello") (3:boolean <- string-equal 1:string-address 2:string-address) ]))) (run 'main) (when (~iso memory*.3 t) (prn "F - 'string-equal'")) (reset) (new-trace "string-equal-empty") (add-code '((function main [ (1:string-address <- new "") (2:string-address <- new "") (3:boolean <- string-equal 1:string-address 2:string-address) ]))) (run 'main) (when (~iso memory*.3 t) (prn "F - 'string-equal' works on empty strings")) (reset) (new-trace "string-equal-compare-with-empty") (add-code '((function main [ (1:string-address <- new "a") (2:string-address <- new "") (3:boolean <- string-equal 1:string-address 2:string-address) ]))) (run 'main) (when (~iso memory*.3 nil) (prn "F - 'string-equal' compares correctly with empty strings")) (reset) (new-trace "string-equal-compares-length") (add-code '((function main [ (1:string-address <- new "a") (2:string-address <- new "ab") (3:boolean <- string-equal 1:string-address 2:string-address) ]))) (run 'main) (when (~iso memory*.3 nil) (prn "F - 'string-equal' handles differing lengths")) (reset) (new-trace "string-equal-compares-initial-element") (add-code '((function main [ (1:string-address <- new "aa") (2:string-address <- new "ba") (3:boolean <- string-equal 1:string-address 2:string-address) ]))) (run 'main) (when (~iso memory*.3 nil) (prn "F - 'string-equal' handles inequal final byte")) (reset) (new-trace "string-equal-compares-final-element") (add-code '((function main [ (1:string-address <- new "ab") (2:string-address <- new "aa") (3:boolean <- string-equal 1:string-address 2:string-address) ]))) (run 'main) (when (~iso memory*.3 nil) (prn "F - 'string-equal' handles inequal final byte")) (reset) (new-trace "string-equal-reflexive") (add-code '((function main [ (1:string-address <- new "ab") (3:boolean <- string-equal 1:string-address 1:string-address) ]))) (run 'main) (when (~iso memory*.3 t) (prn "F - 'string-equal' handles identical pointer")) (reset) (new-trace "strcat") (add-code '((function main [ (1:string-address <- new "hello,") (2:string-address <- new " world!") (3:string-address <- strcat 1:string-address 2:string-address) ]))) ;? (= dump-trace* (obj whitelist '("run"))) ;? 1 (run 'main) (when (~memory-contains-array memory*.3 "hello, world!") (prn "F - 'strcat' concatenates strings")) ;? (quit) ;? 1 (reset) (new-trace "interpolate") (add-code '((function main [ (1:string-address <- new "hello, _!") (2:string-address <- new "abc") (3:string-address <- interpolate 1:string-address 2:string-address) ]))) ;? (= dump-trace* (obj whitelist '("run"))) (run 'main) (when (~memory-contains-array memory*.3 "hello, abc!") (prn "F - 'interpolate' splices strings")) (reset) (new-trace "interpolate-empty") (add-code '((function main [ (1:string-address <- new "hello!") (2:string-address <- new "abc") (3:string-address <- interpolate 1:string-address 2:string-address) ]))) ;? (= dump-trace* (obj whitelist '("run"))) (run 'main) (when (~memory-contains-array memory*.3 "hello!") (prn "F - 'interpolate' without underscore returns template")) (reset) (new-trace "interpolate-at-start") (add-code '((function main [ (1:string-address <- new "_, hello!") (2:string-address <- new "abc") (3:string-address <- interpolate 1:string-address 2:string-address) ]))) ;? (= dump-trace* (obj whitelist '("run"))) (run 'main) (when (~memory-contains-array memory*.3 "abc, hello") (prn "F - 'interpolate' splices strings at start")) (reset) (new-trace "interpolate-at-end") (add-code '((function main [ (1:string-address <- new "hello, _") (2:string-address <- new "abc") (3:string-address <- interpolate 1:string-address 2:string-address) ]))) ;? (= dump-trace* (obj whitelist '("run"))) (run 'main) (when (~memory-contains-array memory*.3 "hello, abc") (prn "F - 'interpolate' splices strings at start")) (reset) (new-trace "interpolate-varargs") (add-code '((function main [ (1:string-address <- new "hello, _, _, and _!") (2:string-address <- new "abc") (3:string-address <- new "def") (4:string-address <- new "ghi") (5:string-address <- interpolate 1:string-address 2:string-address 3:string-address 4:string-address) ]))) ;? (= dump-trace* (obj whitelist '("run"))) ;? (= dump-trace* (obj whitelist '("run" "array-info"))) ;? (set dump-trace*) (run 'main) ;? (quit) ;? (up i 1 (+ 1 (memory* memory*.5)) ;? (prn (memory* (+ memory*.5 i)))) (when (~memory-contains-array memory*.5 "hello, abc, def, and ghi!") (prn "F - 'interpolate' splices in any number of strings")) (reset) (new-trace "string-find-next") (add-code '((function main [ (1:string-address <- new "a/b") (2:integer <- find-next 1:string-address ((#\/ literal)) 0:literal) ]))) (run 'main) (when (~is memory*.2 1) (prn "F - 'find-next' finds first location of a character")) (reset) (new-trace "string-find-next-empty") (add-code '((function main [ (1:string-address <- new "") (2:integer <- find-next 1:string-address ((#\/ literal)) 0:literal) ]))) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (when (~is memory*.2 0) (prn "F - 'find-next' finds first location of a character")) (reset) (new-trace "string-find-next-initial") (add-code '((function main [ (1:string-address <- new "/abc") (2:integer <- find-next 1:string-address ((#\/ literal)) 0:literal) ]))) (run 'main) (when (~is memory*.2 0) (prn "F - 'find-next' handles prefix match")) (reset) (new-trace "string-find-next-final") (add-code '((function main [ (1:string-address <- new "abc/") (2:integer <- find-next 1:string-address ((#\/ literal)) 0:literal) ]))) (run 'main) ;? (prn memory*.2) (when (~is memory*.2 3) (prn "F - 'find-next' handles suffix match")) (reset) (new-trace "string-find-next-missing") (add-code '((function main [ (1:string-address <- new "abc") (2:integer <- find-next 1:string-address ((#\/ literal)) 0:literal) ]))) (run 'main) ;? (prn memory*.2) (when (~is memory*.2 3) (prn "F - 'find-next' handles no match")) (reset) (new-trace "string-find-next-invalid-index") (add-code '((function main [ (1:string-address <- new "abc") (2:integer <- find-next 1:string-address ((#\/ literal)) 4:literal) ]))) ;? (= dump-trace* (obj whitelist '("run"))) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) ;? (prn memory*.2) (when (~is memory*.2 4) (prn "F - 'find-next' skips invalid index (past end of string)")) (reset) (new-trace "string-find-next-first") (add-code '((function main [ (1:string-address <- new "ab/c/") (2:integer <- find-next 1:string-address ((#\/ literal)) 0:literal) ]))) (run 'main) (when (~is memory*.2 2) (prn "F - 'find-next' finds first of multiple options")) (reset) (new-trace "string-find-next-second") (add-code '((function main [ (1:string-address <- new "ab/c/") (2:integer <- find-next 1:string-address ((#\/ literal)) 3:literal) ]))) (run 'main) (when (~is memory*.2 4) (prn "F - 'find-next' finds second of multiple options")) (reset) (new-trace "match-at") (add-code '((function main [ (1:string-address <- new "abc") (2:string-address <- new "ab") (3:boolean <- match-at 1:string-address 2:string-address 0:literal) ]))) (run 'main) (when (~is memory*.3 t) (prn "F - 'match-at' matches substring at given index")) (reset) (new-trace "match-at-reflexive") (add-code '((function main [ (1:string-address <- new "abc") (3:boolean <- match-at 1:string-address 1:string-address 0:literal) ]))) (run 'main) (when (~is memory*.3 t) (prn "F - 'match-at' always matches a string at itself at index 0")) (reset) (new-trace "match-at-outside-bounds") (add-code '((function main [ (1:string-address <- new "abc") (2:string-address <- new "a") (3:boolean <- match-at 1:string-address 2:string-address 4:literal) ]))) (run 'main) (when (~is memory*.3 nil) (prn "F - 'match-at' always fails to match outside the bounds of the text")) (reset) (new-trace "match-at-empty-pattern") (add-code '((function main [ (1:string-address <- new "abc") (2:string-address <- new "") (3:boolean <- match-at 1:string-address 2:string-address 0:literal) ]))) (run 'main) (when (~is memory*.3 t) (prn "F - 'match-at' always matches empty pattern")) (reset) (new-trace "match-at-empty-pattern-outside-bounds") (add-code '((function main [ (1:string-address <- new "abc") (2:string-address <- new "") (3:boolean <- match-at 1:string-address 2:string-address 4:literal) ]))) (run 'main) (when (~is memory*.3 nil) (prn "F - 'match-at' matches empty pattern -- unless index is out of bounds")) (reset) (new-trace "match-at-empty-text") (add-code '((function main [ (1:string-address <- new "") (2:string-address <- new "abc") (3:boolean <- match-at 1:string-address 2:string-address 0:literal) ]))) (run 'main) (when (~is memory*.3 nil) (prn "F - 'match-at' never matches empty text")) (reset) (new-trace "match-at-empty-against-empty") (add-code '((function main [ (1:string-address <- new "") (3:boolean <- match-at 1:string-address 1:string-address 0:literal) ]))) (run 'main) (when (~is memory*.3 t) (prn "F - 'match-at' never matches empty text -- unless pattern is also empty")) (reset) (new-trace "match-at-inside-bounds") (add-code '((function main [ (1:string-address <- new "abc") (2:string-address <- new "bc") (3:boolean <- match-at 1:string-address 2:string-address 1:literal) ]))) (run 'main) (when (~is memory*.3 t) (prn "F - 'match-at' matches inner substring")) (reset) (new-trace "match-at-inside-bounds-2") (add-code '((function main [ (1:string-address <- new "abc") (2:string-address <- new "bc") (3:boolean <- match-at 1:string-address 2:string-address 0:literal) ]))) (run 'main) (when (~is memory*.3 nil) (prn "F - 'match-at' matches inner substring - 2")) (reset) (new-trace "find-substring") (add-code '((function main [ (1:string-address <- new "abc") (2:string-address <- new "bc") (3:integer <- find-substring 1:string-address 2:string-address 0:literal) ]))) (run 'main) ;? (prn memory*.3) ;? 1 (when (~is memory*.3 1) (prn "F - 'find-substring' returns index of match")) (reset) (new-trace "find-substring-2") (add-code '((function main [ (1:string-address <- new "abcd") (2:string-address <- new "bc") (3:integer <- find-substring 1:string-address 2:string-address 1:literal) ]))) (run 'main) (when (~is memory*.3 1) (prn "F - 'find-substring' returns provided index if it matches")) (reset) (new-trace "find-substring-no-match") (add-code '((function main [ (1:string-address <- new "abc") (2:string-address <- new "bd") (3:integer <- find-substring 1:string-address 2:string-address 0:literal) ]))) (run 'main) (when (~is memory*.3 3) (prn "F - 'find-substring' returns out-of-bounds index on no-match")) (reset) (new-trace "find-substring-suffix-match") (add-code '((function main [ (1:string-address <- new "abcd") (2:string-address <- new "cd") (3:integer <- find-substring 1:string-address 2:string-address 0:literal) ]))) (run 'main) (when (~is memory*.3 2) (prn "F - 'find-substring' returns provided index if it matches")) (reset) (new-trace "find-substring-suffix-match-2") (add-code '((function main [ (1:string-address <- new "abcd") (2:string-address <- new "cde") (3:integer <- find-substring 1:string-address 2:string-address 0:literal) ]))) (run 'main) (when (~is memory*.3 4) (prn "F - 'find-substring' returns provided index if it matches")) ;? (quit) ;? 1 (reset) (new-trace "string-split") (add-code '((function main [ (1:string-address <- new "a/b") (2:string-address-array-address <- split 1:string-address ((#\/ literal))) ]))) ;? (set dump-trace*) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (let base memory*.2 ;? (prn base " " memory*.base) (when (or (~is memory*.base 2) ;? (do1 nil prn.111) (~memory-contains-array (memory* (+ base 1)) "a") ;? (do1 nil prn.111) (~memory-contains-array (memory* (+ base 2)) "b")) (prn "F - 'split' cuts string at delimiter"))) (reset) (new-trace "string-split2") (add-code '((function main [ (1:string-address <- new "a/b/c") (2:string-address-array-address <- split 1:string-address ((#\/ literal))) ]))) ;? (set dump-trace*) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (let base memory*.2 ;? (prn base " " memory*.base) (when (or (~is memory*.base 3) ;? (do1 nil prn.111) (~memory-contains-array (memory* (+ base 1)) "a") ;? (do1 nil prn.111) (~memory-contains-array (memory* (+ base 2)) "b") ;? (do1 nil prn.111) (~memory-contains-array (memory* (+ base 3)) "c")) (prn "F - 'split' cuts string at two delimiters"))) (reset) (new-trace "string-split-missing") (add-code '((function main [ (1:string-address <- new "abc") (2:string-address-array-address <- split 1:string-address ((#\/ literal))) ]))) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (let base memory*.2 (when (or (~is memory*.base 1) (~memory-contains-array (memory* (+ base 1)) "abc")) (prn "F - 'split' handles missing delimiter"))) (reset) (new-trace "string-split-empty") (add-code '((function main [ (1:string-address <- new "") (2:string-address-array-address <- split 1:string-address ((#\/ literal))) ]))) ;? (= dump-trace* (obj whitelist '("run"))) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (let base memory*.2 ;? (prn base " " memory*.base) (when (~is memory*.base 0) (prn "F - 'split' handles empty string"))) (reset) (new-trace "string-split-empty-piece") (add-code '((function main [ (1:string-address <- new "a/b//c") (2:string-address-array-address <- split 1:string-address ((#\/ literal))) ]))) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (let base memory*.2 (when (or (~is memory*.base 4) (~memory-contains-array (memory* (+ base 1)) "a") (~memory-contains-array (memory* (+ base 2)) "b") (~memory-contains-array (memory* (+ base 3)) "") (~memory-contains-array (memory* (+ base 4)) "c")) (prn "F - 'split' cuts string at two delimiters"))) ;? (quit) ;? 1 (reset) (new-trace "string-split-first") (add-code '((function main [ (1:string-address <- new "a/b") (2:string-address 3:string-address <- split-first 1:string-address ((#\/ literal))) ]))) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (when (or (~memory-contains-array memory*.2 "a") (~memory-contains-array memory*.3 "b")) (prn "F - 'split-first' cuts string at first occurrence of delimiter")) (reset) (new-trace "string-split-first-at-substring") (add-code '((function main [ (1:string-address <- new "a//b") (2:string-address <- new "//") (3:string-address 4:string-address <- split-first-at-substring 1:string-address 2:string-address) ]))) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) ;? (prn int-canon.memory*) ;? 1 (when (or (~memory-contains-array memory*.3 "a") (~memory-contains-array memory*.4 "b")) (prn "F - 'split-first-at-substring' is like split-first but with a string delimiter")) (reset) (new-trace "string-copy") (add-code '((function main [ (1:string-address <- new "abc") (2:string-address <- string-copy 1:string-address 1:literal 3:literal) ]))) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (when (~memory-contains-array memory*.2 "bc") (prn "F - 'string-copy' returns a copy of a substring")) (reset) (new-trace "string-copy-out-of-bounds") (add-code '((function main [ (1:string-address <- new "abc") (2:string-address <- string-copy 1:string-address 2:literal 4:literal) ]))) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (when (~memory-contains-array memory*.2 "c") (prn "F - 'string-copy' stops at bounds")) (reset) (new-trace "string-copy-out-of-bounds-2") (add-code '((function main [ (1:string-address <- new "abc") (2:string-address <- string-copy 1:string-address 3:literal 3:literal) ]))) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) (when (~memory-contains-array memory*.2 "") (prn "F - 'string-copy' returns empty string when range is out of bounds")) (reset) (new-trace "min") (add-code '((function main [ (1:integer <- min 3:literal 4:literal) ]))) (run 'main) (each routine completed-routines* (aif rep.routine!error (prn "error - " it))) ;? (prn int-canon.memory*) ;? 1 (when (~is memory*.1 3) (prn "F - 'min' returns smaller of two numbers")) ;? (quit) ;? 2 (reset) (new-trace "integer-to-decimal-string") (add-code '((function main [ (1:string-address/raw <- integer-to-decimal-string 34:literal) ]))) ;? (set dump-trace*) ;? (= dump-trace* (obj whitelist '("run"))) (run 'main) (let base memory*.1 (when (~memory-contains-array base "34") (prn "F - converting integer to decimal string"))) (reset) (new-trace "integer-to-decimal-string-zero") (add-code '((function main [ (1:string-address/raw <- integer-to-decimal-string 0:literal) ]))) (run 'main) (let base memory*.1 (when (~memory-contains-array base "0") (prn "F - converting zero to decimal string"))) (reset) (new-trace "integer-to-decimal-string-negative") (add-code '((function main [ (1:string-address/raw <- integer-to-decimal-string -237:literal) ]))) (run 'main) (let base memory*.1 (when (~memory-contains-array base "-237") (prn "F - converting negative integer to decimal string"))) ; fake screen for tests; prints go to a string (reset) (new-trace "fake-screen-empty") (add-code '((function main [ (default-space:space-address <- new space:literal 30:literal/capacity) (screen:terminal-address <- init-fake-terminal 20:literal 10:literal) (5:string-address/raw <- get screen:terminal-address/deref data:offset) ]))) (run 'main) (each routine completed-routines* (awhen rep.routine!error (prn "error - " it))) (when (~memory-contains-array memory*.5 (+ " " " " " " " " " " " " " " " " " " " ")) (prn "F - fake screen starts out with all spaces")) ; fake keyboard for tests; must initialize keys in advance (reset) (new-trace "fake-keyboard") (add-code '((function main [ (default-space:space-address <- new space:literal 30:literal) (s:string-address <- new "foo") (x:keyboard-address <- init-keyboard s:string-address) (1:character-address/raw <- read-key x:keyboard-address) ]))) (run 'main) (when (~is memory*.1 #\f) (prn "F - 'read-key' reads character from provided 'fake keyboard' string")) ; fake keyboard for tests; must initialize keys in advance (reset) (new-trace "fake-keyboard2") (add-code '((function main [ (default-space:space-address <- new space:literal 30:literal) (s:string-address <- new "foo") (x:keyboard-address <- init-keyboard s:string-address) (1:character-address/raw <- read-key x:keyboard-address) (1:character-address/raw <- read-key x:keyboard-address) ]))) (run 'main) (when (~is memory*.1 #\o) (prn "F - 'read-key' advances cursor in provided string")) ; to receive input line by line, run send-keys-buffered-to-stdin (reset) (new-trace "buffer-stdin-until-newline") (add-code '((function main [ (default-space:space-address <- new space:literal 30:literal) (s:string-address <- new "foo") (k:keyboard-address <- init-keyboard s:string-address) (stdin:channel-address <- init-channel 1:literal) (fork send-keys-to-stdin:fn nil:literal/globals nil:literal/limit k:keyboard-address stdin:channel-address) (buffered-stdin:channel-address <- init-channel 1:literal) (r:integer/routine <- fork buffer-lines:fn nil:literal/globals nil:literal/limit stdin:channel-address buffered-stdin:channel-address) (screen:terminal-address <- init-fake-terminal 20:literal 10:literal) (5:string-address/raw <- get screen:terminal-address/deref data:offset) (fork-helper send-prints-to-stdout:fn nil:literal/globals nil:literal/limit screen:terminal-address buffered-stdin:channel-address) (sleep until-routine-done:literal r:integer/routine) ]))) ;? (set dump-trace*) ;? 3 ;? (= dump-trace* (obj whitelist '("schedule" "run"))) ;? 0 (run 'main) ;? (prn int-canon.memory*) ;? 0 (each routine completed-routines* (awhen rep.routine!error (prn "error - " it))) (when (~memory-contains-array memory*.5 (+ " " " " " " " " " " " " " " " " " " " ")) (prn "F - 'buffer-lines' prints nothing until newline is encountered")) ;? (quit) ;? 3 (reset) (new-trace "print-buffered-contents-on-newline") (add-code '((function main [ (default-space:space-address <- new space:literal 30:literal) (s:string-address <- new "foo\nline2") (k:keyboard-address <- init-keyboard s:string-address) (stdin:channel-address <- init-channel 1:literal) (fork send-keys-to-stdin:fn nil:literal/globals nil:literal/limit k:keyboard-address stdin:channel-address) (buffered-stdin:channel-address <- init-channel 1:literal) (r:integer/routine <- fork buffer-lines:fn nil:literal/globals nil:literal/limit stdin:channel-address buffered-stdin:channel-address) (screen:terminal-address <- init-fake-terminal 20:literal 10:literal) (5:string-address/raw <- get screen:terminal-address/deref data:offset) (fork-helper send-prints-to-stdout:fn nil:literal/globals nil:literal/limit screen:terminal-address buffered-stdin:channel-address) (sleep until-routine-done:literal r:integer/routine) ]))) ;? (= dump-trace* (obj whitelist '("schedule" "run"))) ;? 1 (run 'main) (each routine completed-routines* (awhen rep.routine!error (prn "error - " it))) (when (~memory-contains-array memory*.5 (+ "foo\n " " " " " " " " " " " " " " " " " " ")) (prn "F - 'buffer-lines' prints lines to screen")) (reset) (new-trace "print-buffered-contents-right-at-newline") (add-code '((function main [ (default-space:space-address <- new space:literal 30:literal) (s:string-address <- new "foo\n") (k:keyboard-address <- init-keyboard s:string-address) (stdin:channel-address <- init-channel 1:literal) (fork send-keys-to-stdin:fn nil:literal/globals nil:literal/limit k:keyboard-address stdin:channel-address) (buffered-stdin:channel-address <- init-channel 1:literal) (r:integer/routine <- fork buffer-lines:fn nil:literal/globals nil:literal/limit stdin:channel-address buffered-stdin:channel-address) (screen:terminal-address <- init-fake-terminal 20:literal 10:literal) (5:string-address/raw <- get screen:terminal-address/deref data:offset) (fork-helper send-prints-to-stdout:fn nil:literal/globals nil:literal/limit screen:terminal-address buffered-stdin:channel-address) (sleep until-routine-done:literal r:integer/routine) ; hack: give helper some time to finish printing (sleep for-some-cycles:literal 500:literal) ]))) ;? (= dump-trace* (obj whitelist '("schedule" "run"))) ;? 1 (run 'main) (each routine completed-routines* (awhen rep.routine!error (prn "error - " it))) (when (~memory-contains-array memory*.5 (+ "foo\n " " " " " " " " " " " " " " " " " " ")) (prn "F - 'buffer-lines' prints lines to screen immediately on newline")) (reset) (new-trace "buffered-contents-skip-backspace") (add-code '((function main [ (default-space:space-address <- new space:literal 30:literal) (s:string-address <- new "fooa\b\nline2") (k:keyboard-address <- init-keyboard s:string-address) (stdin:channel-address <- init-channel 1:literal) (fork send-keys-to-stdin:fn nil:literal/globals nil:literal/limit k:keyboard-address stdin:channel-address) (buffered-stdin:channel-address <- init-channel 1:literal) (r:integer/routine <- fork buffer-lines:fn nil:literal/globals nil:literal/limit stdin:channel-address buffered-stdin:channel-address) (screen:terminal-address <- init-fake-terminal 20:literal 10:literal) (5:string-address/raw <- get screen:terminal-address/deref data:offset) (fork-helper send-prints-to-stdout:fn nil:literal/globals nil:literal/limit screen:terminal-address buffered-stdin:channel-address) (sleep until-routine-done:literal r:integer/routine) ]))) ;? (= dump-trace* (obj whitelist '("schedule" "run"))) ;? 1 (run 'main) (each routine completed-routines* (awhen rep.routine!error (prn "error - " it))) (when (~memory-contains-array memory*.5 (+ "foo\n " " " " " " " " " " " " " " " " " " ")) (prn "F - 'buffer-lines' handles backspace")) (reset) (new-trace "buffered-contents-ignore-excess-backspace") (add-code '((function main [ (default-space:space-address <- new space:literal 30:literal) (s:string-address <- new "a\b\bfoo\n") (k:keyboard-address <- init-keyboard s:string-address) (stdin:channel-address <- init-channel 1:literal) (fork send-keys-to-stdin:fn nil:literal/globals nil:literal/limit k:keyboard-address stdin:channel-address) (buffered-stdin:channel-address <- init-channel 1:literal) (r:integer/routine <- fork buffer-lines:fn nil:literal/globals nil:literal/limit stdin:channel-address buffered-stdin:channel-address) (screen:terminal-address <- init-fake-terminal 20:literal 10:literal) (5:string-address/raw <- get screen:terminal-address/deref data:offset) (fork-helper send-prints-to-stdout:fn nil:literal/globals nil:literal/limit screen:terminal-address buffered-stdin:channel-address) (sleep until-routine-done:literal r:integer/routine) ; hack: give helper some time to finish printing (sleep for-some-cycles:literal 500:literal) ]))) ;? (= dump-trace* (obj whitelist '("schedule" "run"))) ;? 1 (run 'main) (each routine completed-routines* (awhen rep.routine!error (prn "error - " it))) ;? (prn memory*.5) ;? 1 (when (~memory-contains-array memory*.5 (+ "foo\n " " " " " " " " " " " " " " " " " " ")) (prn "F - 'buffer-lines' ignores backspace when there's nothing to backspace over")) ) ; section 100 (reset) (new-trace "parse-and-record") (add-code '((and-record foo [ x:string y:integer z:boolean ]))) (when (~iso type*!foo (obj size 3 and-record t elems '((string) (integer) (boolean)) fields '(x y z))) (prn "F - 'add-code' can add new and-records")) ;; unit tests for various helpers ; tokenize-args (prn "== tokenize-args") (assert:iso '((a b) (c d)) (tokenize-arg 'a:b/c:d)) ; numbers are not symbols (assert:iso '((a b) (1 d)) (tokenize-arg 'a:b/1:d)) ; special symbols are skipped (assert:iso '<- (tokenize-arg '<-)) (assert:iso '_ (tokenize-arg '_)) ; idempotent (assert:iso (tokenize-arg:tokenize-arg 'a:b/c:d) (tokenize-arg 'a:b/c:d)) ; support labels (assert:iso '((((default-space space-address)) <- ((new)) ((space literal)) ((30 literal))) foo) (tokenize-args '((default-space:space-address <- new space:literal 30:literal) foo))) ; support braces (assert:iso '((((default-space space-address)) <- ((new)) ((space literal)) ((30 literal))) foo { begin bar (((a b)) <- ((op)) ((c d)) ((e f))) }) (tokenize-args '((default-space:space-address <- new space:literal 30:literal) foo { begin bar (a:b <- op c:d e:f) }))) ; space (prn "== space") (reset) (when (~iso 0 (space '((4 integer)))) (prn "F - 'space' is 0 by default")) (when (~iso 1 (space '((4 integer) (space 1)))) (prn "F - 'space' picks up space when available")) (when (~iso 'global (space '((4 integer) (space global)))) (prn "F - 'space' understands routine-global space")) ; absolutize (prn "== absolutize") (reset) (when (~iso '((4 integer)) (absolutize '((4 integer)))) (prn "F - 'absolutize' works without routine")) (= routine* make-routine!foo) (when (~iso '((4 integer)) (absolutize '((4 integer)))) (prn "F - 'absolutize' works without default-space")) (= rep.routine*!call-stack.0!default-space 10) (= memory*.10 5) ; bounds check for default-space (when (~iso '((15 integer) (raw)) (absolutize '((4 integer)))) (prn "F - 'absolutize' works with default-space")) (absolutize '((5 integer))) (when (~posmatch "no room" rep.routine*!error) (prn "F - 'absolutize' checks against default-space bounds")) (when (~iso '((_ integer)) (absolutize '((_ integer)))) (prn "F - 'absolutize' passes dummy args right through")) (when (~iso '((default-space integer)) (absolutize '((default-space integer)))) (prn "F - 'absolutize' passes 'default-space' right through")) (= memory*.20 5) ; pretend array (= rep.routine*!globals 20) ; provide it to routine global (when (~iso '((22 integer) (raw)) (absolutize '((1 integer) (space global)))) (prn "F - 'absolutize' handles variables in the global space")) ; deref (prn "== deref") (reset) (= memory*.3 4) (when (~iso '((4 integer)) (deref '((3 integer-address) (deref)))) (prn "F - 'deref' handles simple addresses")) (when (~iso '((4 integer) (deref)) (deref '((3 integer-address) (deref) (deref)))) (prn "F - 'deref' deletes just one deref")) (= memory*.4 5) (when (~iso '((5 integer)) (deref:deref '((3 integer-address-address) (deref) (deref)))) (prn "F - 'deref' can be chained")) (when (~iso '((5 integer) (foo)) (deref:deref '((3 integer-address-address) (deref) (foo) (deref)))) (prn "F - 'deref' skips junk")) ; addr (prn "== addr") (reset) (= routine* nil) ;? (prn 111) (when (~is 4 (addr '((4 integer)))) (prn "F - directly addressed operands are their own address")) ;? (quit) (when (~is 4 (addr '((4 integer-address)))) (prn "F - directly addressed operands are their own address - 2")) (when (~is 4 (addr '((4 literal)))) (prn "F - 'addr' doesn't understand literals")) ;? (prn 201) (= memory*.4 23) ;? (prn 202) (when (~is 23 (addr '((4 integer-address) (deref)))) (prn "F - 'addr' works with indirectly-addressed 'deref'")) ;? (quit) (= memory*.3 4) (when (~is 23 (addr '((3 integer-address-address) (deref) (deref)))) (prn "F - 'addr' works with multiple 'deref'")) (= routine* make-routine!foo) (when (~is 4 (addr '((4 integer)))) (prn "F - directly addressed operands are their own address inside routines")) (when (~is 4 (addr '((4 integer-address)))) (prn "F - directly addressed operands are their own address inside routines - 2")) (when (~is 4 (addr '((4 literal)))) (prn "F - 'addr' doesn't understand literals inside routines")) (= memory*.4 23) (when (~is 23 (addr '((4 integer-address) (deref)))) (prn "F - 'addr' works with indirectly-addressed 'deref' inside routines")) ;? (prn 301) (= rep.routine*!call-stack.0!default-space 10) ;? (prn 302) (= memory*.10 5) ; bounds check for default-space ;? (prn 303) (when (~is 15 (addr '((4 integer)))) (prn "F - directly addressed operands in routines add default-space")) ;? (quit) (when (~is 15 (addr '((4 integer-address)))) (prn "F - directly addressed operands in routines add default-space - 2")) (when (~is 15 (addr '((4 literal)))) (prn "F - 'addr' doesn't understand literals")) (= memory*.15 23) (when (~is 23 (addr '((4 integer-address) (deref)))) (prn "F - 'addr' adds default-space before 'deref', not after")) ;? (quit) ; array-len (prn "== array-len") (reset) (= memory*.35 4) (when (~is 4 (array-len '((35 integer-boolean-pair-array)))) (prn "F - 'array-len'")) (= memory*.34 35) (when (~is 4 (array-len '((34 integer-boolean-pair-array-address) (deref)))) (prn "F - 'array-len'")) ;? (quit) ; sizeof (prn "== sizeof") (reset) ;? (set dump-trace*) ;? (prn 401) (when (~is 1 (sizeof '((_ integer)))) (prn "F - 'sizeof' works on primitives")) (when (~is 1 (sizeof '((_ integer-address)))) (prn "F - 'sizeof' works on addresses")) (when (~is 2 (sizeof '((_ integer-boolean-pair)))) (prn "F - 'sizeof' works on and-records")) (when (~is 3 (sizeof '((_ integer-point-pair)))) (prn "F - 'sizeof' works on and-records with and-record fields")) ;? (prn 410) (when (~is 1 (sizeof '((34 integer)))) (prn "F - 'sizeof' works on primitive operands")) (when (~is 1 (sizeof '((34 integer-address)))) (prn "F - 'sizeof' works on address operands")) (when (~is 2 (sizeof '((34 integer-boolean-pair)))) (prn "F - 'sizeof' works on and-record operands")) (when (~is 3 (sizeof '((34 integer-point-pair)))) (prn "F - 'sizeof' works on and-record operands with and-record fields")) (when (~is 2 (sizeof '((34 integer-boolean-pair-address) (deref)))) (prn "F - 'sizeof' works on pointers to and-records")) (= memory*.35 4) ; size of array (= memory*.34 35) ;? (= dump-trace* (obj whitelist '("sizeof" "array-len"))) (when (~is 9 (sizeof '((34 integer-boolean-pair-array-address) (deref)))) (prn "F - 'sizeof' works on pointers to arrays")) ;? (quit) ;? (prn 420) (= memory*.4 23) (when (~is 24 (sizeof '((4 integer-array)))) (prn "F - 'sizeof' reads array lengths from memory")) (= memory*.3 4) (when (~is 24 (sizeof '((3 integer-array-address) (deref)))) (prn "F - 'sizeof' handles pointers to arrays")) (= memory*.15 34) (= routine* make-routine!foo) (when (~is 24 (sizeof '((4 integer-array)))) (prn "F - 'sizeof' reads array lengths from memory inside routines")) (= rep.routine*!call-stack.0!default-space 10) (= memory*.10 5) ; bounds check for default-space (when (~is 35 (sizeof '((4 integer-array)))) (prn "F - 'sizeof' reads array lengths from memory using default-space")) (= memory*.35 4) ; size of array (= memory*.15 35) ;? (= dump-trace* (obj whitelist '("sizeof"))) (aif rep.routine*!error (prn "error - " it)) (when (~is 9 (sizeof '((4 integer-boolean-pair-array-address) (deref)))) (prn "F - 'sizeof' works on pointers to arrays using default-space")) ;? (quit) ; m (prn "== m") (reset) (when (~is 4 (m '((4 literal)))) (prn "F - 'm' avoids reading memory for literals")) (when (~is 4 (m '((4 offset)))) (prn "F - 'm' avoids reading memory for offsets")) (= memory*.4 34) (when (~is 34 (m '((4 integer)))) (prn "F - 'm' reads memory for simple types")) (= memory*.3 4) (when (~is 34 (m '((3 integer-address) (deref)))) (prn "F - 'm' redirects addresses")) (= memory*.2 3) (when (~is 34 (m '((2 integer-address-address) (deref) (deref)))) (prn "F - 'm' multiply redirects addresses")) (when (~iso (annotate 'record '(34 nil)) (m '((4 integer-boolean-pair)))) (prn "F - 'm' supports compound records")) (= memory*.5 35) (= memory*.6 36) (when (~iso (annotate 'record '(34 35 36)) (m '((4 integer-point-pair)))) (prn "F - 'm' supports records with compound fields")) (when (~iso (annotate 'record '(34 35 36)) (m '((3 integer-point-pair-address) (deref)))) (prn "F - 'm' supports indirect access to records")) (= memory*.4 2) (when (~iso (annotate 'record '(2 35 36)) (m '((4 integer-array)))) (prn "F - 'm' supports access to arrays")) (when (~iso (annotate 'record '(2 35 36)) (m '((3 integer-array-address) (deref)))) (prn "F - 'm' supports indirect access to arrays")) (= routine* make-routine!foo) (= memory*.10 5) ; fake array (= memory*.12 34) (= rep.routine*!globals 10) (when (~iso 34 (m '((1 integer) (space global)))) (prn "F - 'm' supports access to per-routine globals")) ; setm (prn "== setm") (reset) (setm '((4 integer)) 34) (when (~is 34 memory*.4) (prn "F - 'setm' writes primitives to memory")) (setm '((3 integer-address)) 4) (when (~is 4 memory*.3) (prn "F - 'setm' writes addresses to memory")) (setm '((3 integer-address) (deref)) 35) (when (~is 35 memory*.4) (prn "F - 'setm' redirects writes")) (= memory*.2 3) (setm '((2 integer-address-address) (deref) (deref)) 36) (when (~is 36 memory*.4) (prn "F - 'setm' multiply redirects writes")) ;? (prn 505) (setm '((4 integer-integer-pair)) (annotate 'record '(23 24))) (when (~memory-contains 4 '(23 24)) (prn "F - 'setm' writes compound records")) (assert (is memory*.7 nil)) ;? (prn 506) (setm '((7 integer-point-pair)) (annotate 'record '(23 24 25))) (when (~memory-contains 7 '(23 24 25)) (prn "F - 'setm' writes records with compound fields")) (= routine* make-routine!foo) (setm '((4 integer-point-pair)) (annotate 'record '(33 34))) (when (~posmatch "incorrect size" rep.routine*!error) (prn "F - 'setm' checks size of target")) (wipe routine*) (setm '((3 integer-point-pair-address) (deref)) (annotate 'record '(43 44 45))) (when (~memory-contains 4 '(43 44 45)) (prn "F - 'setm' supports indirect writes to records")) (setm '((2 integer-point-pair-address-address) (deref) (deref)) (annotate 'record '(53 54 55))) (when (~memory-contains 4 '(53 54 55)) (prn "F - 'setm' supports multiply indirect writes to records")) (setm '((4 integer-array)) (annotate 'record '(2 31 32))) (when (~memory-contains 4 '(2 31 32)) (prn "F - 'setm' writes arrays")) (setm '((3 integer-array-address) (deref)) (annotate 'record '(2 41 42))) (when (~memory-contains 4 '(2 41 42)) (prn "F - 'setm' supports indirect writes to arrays")) (= routine* make-routine!foo) (setm '((4 integer-array)) (annotate 'record '(2 31 32 33))) (when (~posmatch "invalid array" rep.routine*!error) (prn "F - 'setm' checks that array written is well-formed")) (= routine* make-routine!foo) ;? (prn 111) ;? (= dump-trace* (obj whitelist '("sizeof" "mem"))) (setm '((4 integer-boolean-pair-array)) (annotate 'record '(2 31 nil 32 nil 33))) (when (~posmatch "invalid array" rep.routine*!error) (prn "F - 'setm' checks that array of records is well-formed")) (= routine* make-routine!foo) ;? (prn 222) (setm '((4 integer-boolean-pair-array)) (annotate 'record '(2 31 nil 32 nil))) (when (posmatch "invalid array" rep.routine*!error) (prn "F - 'setm' checks that array of records is well-formed - 2")) (wipe routine*) (reset) ; end file with this to persist the trace for the final test