pkgsrc

kpriv/pkgsrc

Fork 0

Commit graph

Author	SHA1	Message	Date
epg	1fdb8a39b2	Version 1.1.4 (1 April 2005, from /branches/1.1.x) http://svn.collab.net/repos/svn/tags/1.1.4 User-visible-changes: - Client: * fixed: win32 not ignoring versioned symlinks (issue #2173) * fixed: 'svn merge' can cause broken working copy (issue #2222) * fixed: 'svn commit' fails when schedule-delete dir has local mod (r11980) * fixed: 'svn st -u nonexistent_file' segfault (issue #2127) * fixed: 'svn cp wc wc' utf8 conversion error (r13111) * fixed: confusing error message about "wc not locked" (issue #2174) * many translation updates for localized client messages - Server: * fixed: nasty (though unusual) performance bug in FSFS commits (r13222-3) * fixed: FSFS memory leak when auto-merging large tree (r13193) * fixed: FSFS memory leak in 'svnadmin hotcopy' (r13218, 13465, 13468) * fixed: FSFS segfault when encountering empty data reps (r13683) * fixed: two dataloss bugs in svndumpfilter (r12630, r12636) * fixed: wasteful memory usage in svndumpfilter (r12637, r12640) * fixed: mod_dav_svn segfaults when client sends bogus paths (issue #2199) - Both: * fixed: (win32) retry file operation if sharing violation (r12983, r12986) Developer-visible-changes: * add SWIG 1.3.24 and .25 compatibility (r12551, r12717-9, r12722, r13504) * make mailer.py work on win32 (r12499, r12542, r12670) * fixed: JavaHL run-time link error (r12576), path/url cleanups (r13090) * fixed: python bindings log_receiver failure with SWIG 1.3.24 (r13487) * build system tweaks: add install dependencies for fs & fs_base (r11050)	2005-04-05 01:58:05 +00:00
epg	760b89670a	We used to avoid building all of Subversion again when building sub-packages (p5-subversion, py-subversion, and ap2-subversion) by hand-hacking build-outputs.mk, which was a big pain. Consequently, since the 1.0.7 update, pkgsrc has re-built the whole thing for each subpackage. This is annoying and makes p5-subversion not work with pkgviews on Darwin. Now, we'll fix the problem by patching gen-make.py, which generates build-outputs.mk. I may even be able to commit this change upstream. devel/subversion-base/Makefile: Generate build-outputs.mk hack. devel/p5-subversion/Makefile: devel/py-subversion/Makefile: www/ap2-subversion/Makefile: Replace build-outputs.mk in post-patch.	2004-11-27 02:35:52 +00:00

Author

SHA1

Message

Date

epg

1fdb8a39b2

Version 1.1.4

(1 April 2005, from /branches/1.1.x)
http://svn.collab.net/repos/svn/tags/1.1.4

 User-visible-changes:
  - Client:
    * fixed: win32 not ignoring versioned symlinks (issue #2173)
    * fixed: 'svn merge' can cause broken working copy (issue #2222)
    * fixed: 'svn commit' fails when schedule-delete dir has local mod (r11980)
    * fixed: 'svn st -u nonexistent_file' segfault (issue #2127)
    * fixed: 'svn cp wc wc' utf8 conversion error (r13111)
    * fixed: confusing error message about "wc not locked" (issue #2174)
    * many translation updates for localized client messages

  - Server:
    * fixed: nasty (though unusual) performance bug in FSFS commits (r13222-3)
    * fixed: FSFS memory leak when auto-merging large tree (r13193)
    * fixed: FSFS memory leak in 'svnadmin hotcopy' (r13218, 13465, 13468)
    * fixed: FSFS segfault when encountering empty data reps (r13683)
    * fixed: two dataloss bugs in svndumpfilter (r12630, r12636)
    * fixed: wasteful memory usage in svndumpfilter (r12637, r12640)
    * fixed: mod_dav_svn segfaults when client sends bogus paths (issue #2199)

  - Both:
    * fixed: (win32) retry file operation if sharing violation (r12983, r12986)

 Developer-visible-changes:
 * add SWIG 1.3.24 and .25 compatibility (r12551, r12717-9, r12722, r13504)
 * make mailer.py work on win32 (r12499, r12542, r12670)
 * fixed: JavaHL run-time link error (r12576), path/url cleanups (r13090)
 * fixed: python bindings log_receiver failure with SWIG 1.3.24 (r13487)
 * build system tweaks: add install dependencies for fs & fs_base (r11050)

2005-04-05 01:58:05 +00:00

epg

760b89670a

We used to avoid building all of Subversion again when building

sub-packages (p5-subversion, py-subversion, and ap2-subversion) by
hand-hacking build-outputs.mk, which was a big pain.  Consequently,
since the 1.0.7 update, pkgsrc has re-built the whole thing for each
subpackage.  This is annoying and makes p5-subversion not work with
pkgviews on Darwin.

Now, we'll fix the problem by patching gen-make.py, which generates
build-outputs.mk.  I may even be able to commit this change upstream.

devel/subversion-base/Makefile:
    Generate build-outputs.mk hack.

devel/p5-subversion/Makefile:
devel/py-subversion/Makefile:
www/ap2-subversion/Makefile:
    Replace build-outputs.mk in post-patch.

2004-11-27 02:35:52 +00:00

2 commits