summaryrefslogtreecommitdiff
path: root/ChangeLog
diff options
context:
space:
mode:
authorPaul Smith <psmith@gnu.org>1999-09-03 22:46:28 +0000
committerPaul Smith <psmith@gnu.org>1999-09-03 22:46:28 +0000
commite49b2f15c9c875f7db16064834ffac123f48b612 (patch)
treed38ac0ac88ba4b64006ecbfad092d1446eef47d9 /ChangeLog
parent10fee1c03c9dcf887314c04612224aeadcba6dd7 (diff)
downloadgunmake-e49b2f15c9c875f7db16064834ffac123f48b612.tar.gz
* Removed a memory stomp (reading freed memory).
* Fixed some infinite recursion on rebuilt makefiles bugs.
Diffstat (limited to 'ChangeLog')
-rw-r--r--ChangeLog21
1 files changed, 21 insertions, 0 deletions
diff --git a/ChangeLog b/ChangeLog
index 3b80863..4e69369 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,24 @@
+1999-09-03 Paul D. Smith <psmith@gnu.org>
+
+ * remake.c (notice_finished_file): Always set mtime_before_update
+ if it's not been set, not just if we ran some rules. Otherwise we
+ may have a situation where a target's prerequisite was rebuilt but
+ not changed, so this target's rules weren't run, then
+ update_goal_chain() sees mtime_before_update != last_mtime and
+ thinks that the top-level target changed when it really didn't.
+ This can cause an infinite loop when remaking makefiles.
+ (update_goal_chain): If we get back to the top and we don't know
+ what the goal's last_mtime was, find it now. We need to know so
+ we can compare it to mtime_before_update later (this is only
+ crucial when remaking makefiles--should we only do it then?)
+
+1999-09-02 Paul D. Smith <psmith@gnu.org>
+
+ * read.c (read_makefile): If "override" appears as the first
+ prerequisite, look further to ensure this is really a
+ target-specific variable definition, and not just some
+ prerequisite named "override".
+
1999-09-01 Paul D. Smith <psmith@gnu.org>
* function.c (IS_PATHSEP) [WINDOWS32]: Allow backslash separators