about summary refs log tree commit diff stats
diff options
context:
space:
mode:
authorKartik Agaram <vc@akkartik.com>2018-05-07 20:22:56 -0700
committerKartik Agaram <vc@akkartik.com>2018-05-07 20:22:56 -0700
commita809b975108e638fe99a295ce20327e6749499e7 (patch)
treed040929aa0fe9177380cddc5d5d149a478663752
parent1233f3cd9668ea34c9de680a1162734da3c71d79 (diff)
downloadmu-a809b975108e638fe99a295ce20327e6749499e7.tar.gz
4238
-rwxr-xr-xbuild47
1 files changed, 4 insertions, 3 deletions
diff --git a/build4 b/build4
index 6a7c40a8..132d9138 100755
--- a/build4
+++ b/build4
@@ -10,9 +10,10 @@
 #      'upstream', build2. And even then, diff gets confused.
 #   5. There's a mechanical difficulty: we use mktemp to reliably create
 #      temporary filenames, which has the side effect of also creating the
-#      files. So zero-size files are considered equivalent to empty files.
-#      When some commands have trouble with this (e.g. ar) we need to delete
-#      the empty file, which can expose us to a race condition wrt mktemp.
+#      files. So zero-size files are considered equivalent to non-existent
+#      files. When some commands have trouble with this (e.g. ar) we need to
+#      delete the empty file, which can expose us to a race condition wrt
+#      mktemp.
 
 set -e  # stop immediately on error