CVS update [cvs1-11-x-branch]: /ccvs/src/

[email protected] 20 Apr 2005 21:08:11 -0000
Newsgroups gmane.comp.version-control.cvs.cvs
Message-ID <[email protected]>
Tag: cvs1-11-x-branch
User: dprice  
Date: 05/04/20 14:08:11

Modified:
 /ccvs/src/
  ChangeLog, sanity.sh

Log:
 * sanity.sh (tests): Add rcs4.
 (rcs5): Add comments.

File Changes:

Directory: /ccvs/src/
=====================

File [changed]: ChangeLog
Url: https://ccvs.cvshome.org/source/browse/ccvs/src/ChangeLog?r1=1.2336.2.361&r2=1.2336.2.362
Delta lines:  +6 -1
-------------------
--- ChangeLog	20 Apr 2005 20:40:42 -0000	1.2336.2.361
+++ ChangeLog	20 Apr 2005 21:07:59 -0000	1.2336.2.362
@@ -1,4 +1,9 @@
-2005-03-20  Derek Price  <[email protected]>
+2005-04-20  Derek Price  <[email protected]>
+
+	* sanity.sh (tests): Add rcs4.
+	(rcs5): Add comments.
+
+2005-04-20  Derek Price  <[email protected]>
 
 	* rcs.c (expand_keywords): Avoid buffer overflow.
 	(Original patch from Stewart Brodie <[email protected]>.)

File [changed]: sanity.sh
Url: https://ccvs.cvshome.org/source/browse/ccvs/src/sanity.sh?r1=1.752.2.165&r2=1.752.2.166
Delta lines:  +12 -1
--------------------
--- sanity.sh	20 Apr 2005 20:40:43 -0000	1.752.2.165
+++ sanity.sh	20 Apr 2005 21:07:59 -0000	1.752.2.166
@@ -1108,7 +1108,7 @@
 	tests="${tests} serverpatch log log2 logopt ann ann-id"
 	# Repository Storage (RCS file format, CVS lock files, creating
 	# a repository without "cvs init", &c).
-	tests="${tests} crerepos rcs rcs2 rcs3 rcs5"
+	tests="${tests} crerepos rcs rcs2 rcs3 rcs4 rcs5"
 	tests="$tests lockfiles backuprecover"
 	tests="${tests} sshstdio"
 	# More history browsing, &c.
@@ -20421,6 +20421,17 @@
 
 
 	rcs5)
+	  # Some tests of the $Log keyword and log message without a trailing
+	  # EOL.  This used to look ugly and, in the worst case, could cause
+	  # a seg fault due to a buffer overflow.
+	  #
+	  # Note that it should not be possible to create this situation via a
+	  # CVS server (and any client), since the server itself inserts the
+	  # trailing EOL onto log messages that are missing one.  Still, we
+	  # shouldn't segfault due to a corrupt RCS file and I think that a log
+	  # message without the trailing EOL doesn't actually violate the RCS
+	  # spec, though it doesn't appear to be possible to create such a log
+	  # message using RCS 5.7.
 	  mkdir $CVSROOT_DIRNAME/rcs5
 	  cat <<\EOF >$CVSROOT_DIRNAME/rcs5/file1,v
 head 1.1;