summaryrefslogtreecommitdiff
path: root/ChangeLog
diff options
context:
space:
mode:
Diffstat (limited to 'ChangeLog')
-rw-r--r--ChangeLog65
1 files changed, 65 insertions, 0 deletions
diff --git a/ChangeLog b/ChangeLog
index 1a9afd9..f4e1592 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,25 @@
+2000-03-27 Paul D. Smith <psmith@gnu.org>
+
+ * remake.c (start_updating, finish_updating, is_updating): Fix
+ PR/1671; circular dependencies in double-colon rules are not
+ diagnosed. These macros set the updating flag in the root
+ double-colon file instead of the current one, if it's part of a
+ double-colon list. This solution provided by Tim Magill
+ <magill@gate.net>; I just changed the macro names :).
+ (update_file_1): Call them.
+ (check_dep): Call them.
+
+ The change to not automatically evaluate the $(call ...)
+ function's arguments breaks recursive use of call. Although using
+ $(if ...) and $(foreach ...) in $(call ...) macros is important,
+ the error conditions generated are simply to obscure for me to
+ feel comfortable with. If a method is devised to get both
+ working, we'll revisit. For now, remove this change.
+
+ * function.c (function_table): Turn on the expand bit for func_call.
+ (func_call): Don't expand arguments for builtin functions; that
+ will have already been done.
+
2000-03-26 Paul D. Smith <psmith@gnu.org>
* file.c (remove_intermediates): Never remove targets explicitly
@@ -23,6 +45,48 @@
* make.h: Ditto.
Reported by Marco Franzen <Marco.Franzen@Thyron.com>.
+2000-03-08 Tim Magill <magill@gate.net>
+
+ * remake.c (update_file): Return the exit status of the pruned
+ file when pruning, not just 0. Fixes PR/1634.
+
+2000-02-24 Paul D. Smith <psmith@gnu.org>
+
+ * configure.in: Close a minor potential security hole; if you're
+ reading makefiles from stdin (who does that?) you could run into a
+ race condition with the temp file using mktemp() or tmpnam(). Add
+ a check for mkstemp() and fdopen().
+ * main.c (open_tmpfile): New function to open a temporary file.
+ If we have mkstemp() (and fdopen()), use that. If not use
+ mktemp() or tmpnam(). If we have fdopen(), use open() to open the
+ file O_CREAT|O_EXCL. If not, fall back to normal fopen() (insecure).
+ (main): Call it.
+ * job.c (child_execute_job) [VMS]: Call it.
+
+ * variable.c (lookup_variable): If we find a variable which is
+ being expanded, then note it but keep looking through the rest of
+ the set list to see if we can find one that isn't. If we do,
+ return that. If we don't, return the original. Fix for PR/1610.
+
+ While implementing this I realized that it also solves PR/1380 in
+ a much more elegant way. I don't know what I was smoking before.
+ So, remove the hackage surrounding the original fix for that (see
+ below). Change this function back to lookup_variable and remove
+ the extra setlist argument.
+ * variable.h (recursively_expand_setlist): Remove the macro,
+ rename the prototype, and remove the extra setlist argument.
+ (lookup_variable): Ditto.
+ * expand.c (recursively_expand): Rename and remove the extra
+ setlist argument.
+ (reference_variable): Use lookup_variable() again.
+ (allocated_variable_append): Remove the extra setlist argument.
+
+2000-02-21 Paul D. Smith <psmith@gnu.org>
+
+ * README.template: A few updates.
+
+ * i18n/de.po: New version from the German translation team.
+
2000-02-09 Paul D. Smith <psmith@gnu.org>
* Version 3.78.91 released.
@@ -42,6 +106,7 @@
implementing expansion of these references separately from the
"normal" expansion, say, instead of using the same codepath.
Actually, it's already "worse enough" :-/.
+ Fix for PR/1380.
* variable.h (recursively_expand_setlist): Rename
recursively_expand to add a struct variable_set_list argument, and