* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * THIS RELEASE STREAM IS OPEN TO BUG FIXES. * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * This file tracks the status of releases in the 1.6.x line. See "Stabilizing and maintaining releases" in trunk/www/hacking.html for details on how release lines and voting work, what kinds of bugs can delay a release, etc. Status of 1.6.0: ================ Candidate changes: * r36559 Fix warning in Python 2.6 for ctypes-python bindings Votes: +1: djames, arfrever +1: gstein (for 1.6.1; this can wait) * r36093, r36292, r36406 Improve some messages. Votes: +1: arfrever +1: gstein, rhuijben (for 1.6.1; this can wait) * r36209 Update paths to svnmucc. Votes: +1: arfrever +1: gstein, rhuijben (for 1.6.1; this can wait) * r36388 Recommend Neon 0.28.4. Votes: +1: arfrever, steveking (already using neon 0.28.4 in TSVN) +1: gstein, rhuijben (for 1.6.1; this can wait) Candidate if a new tarball is constructed (otherwise for 1.6.1): * r36049-r36051, r36131 Resolve the SVN_ERR_MALFUNCTION() reproducible on Windows with svn ci "C:/Program FIles" C:/Windows by handling parent directories correctly. (r36050 and r36050 are typo fixes; r36131 some extra cleanup) Justification: This malfunction makes update_tests.py 32 abort on Windows. Which in turn delays the Windows automated testrun until a user confirms. (This doesn't fix this testcase; it just handles this part of the problem) Votes: +1: rhuijben, pburba, steveking +1: gstein (for a new CODE tarball, not a doc tarball) * r36159 Fix proxying of LOCK and UNLOCK requests via WebDAV proxies. Justification: With the fix, each slave server tries to field the lock/unlock, resulting in competing ideas of what is and isn't locked. Votes: +1: cmpilato, rhuijben, arfrever +1: gstein (for 1.6.1) * r36238 Fix build of RHEL5 RPM packages. Justification: Won't build properly without these changes. Not a release blocker for 1.6.0 but would be nice if the train is still in the station. Otherwise please feel free to move to 1.6.1. Votes: +1: dws, arfrever +0: gstein (for 1.6.0) +1: gstein (for 1.6.1) * r36252-36253 (on branch 1.6.x-r36252) Fix valgrind UMRs. Justification: UMRs are undefined behaviour Votes: +1: philip, rhuijben, stsp, arfrever +1: gstein (for a new CODE tarball, not a doc tarball) * r36302 Fix a small problem in the loggy upgrade process. Some v10 working copies have .svn/format, and some don't. The *intent* is to remove the file if present, so make the code match. Justification: If change-svn-wc-format.py is used to downgrade a working copy from 1.6 to 1.5 format, the script does not add the format file, rendering 1.6 clients unable to update the working copy again: "svn: Can't remove file '.svn/format': No such file or directory" Since the presence of that file has been deprecated for ages, let's just apply this easy fix to 1.6 instead of tweaking the script. Votes: +1: rhuijben, stsp, arfrever -0: gstein (for 1.6.0 since it isn't "needed" except for when somebody uses a script which I conceptually disagree with) +1: gstein (for 1.6.1) * r36213 Trivial help output fix. Justification: Errors in documentation are bad, obvious fix. Notes: Already merged to 1.6.x by Arfrever in r36214, but reverted in 36322 because it was not formally voted on. Votes: +1: stsp, arfrever +1: gstein (for 1.6.1; it can wait) * r36613, r36615, r36631, r36641 Avoid potential segfault when dealing with subtree mergeinfo. Justification: One user reported a segfault due to the way, prior to this change, that we set children_with_mergeinfo elements to NULL rather than removing the elements from the array, see http://svn.haxx.se/dev/archive-2009-03/0382.shtml. Notes: While segfaults are serious, this one is not sufficient by itself to hold up 1.6.0 given that it is not a regression from 1.5.x, no one has been able to produce a reproduction recipe, and only one user has reported it (and they were able to work around it). The fix itself is r36613, r36631, and r36641. r36615 is a variable renaming change that is necessary to avoid a conflict. Votes: +1: pburba Status of 1.6.1: ================ Candidate changes: * r36178 External merge-tool improvements. Justification: Makes automated merging easier, and also fixes a bug where a file which should be marked conflicted was consired sucessfully merged because the tool's exit status was not properly evaluated. Notes: The bug is only triggered when using an external merge tool with the --accept=launch option. I don't think we need to hold 1.6.0 on this. Votes: -0: rhuijben (Looks like a feature enhancement for a patch release) +1: stsp (I'd say it's a fix for a feature we already support (merge under control of external non-interactive tool), albeit the feature is currently so buggy that you can't use it.) * r36256 Fix potentially awful performance of --ignore-ancestry merges. Justification: A very simple change which avoids an unneeded walk of the entire merge target. For large WCs the performance improvment can be dramatic. See http://svn.haxx.se/dev/archive-2009-03/0067.shtml. Votes: +1: pburba, rhuijben, gstein, arfrever * r36444 Optimize merge performance when dealing with subtree mergeinfo. Justification: Another simple change which, in many common situations, avoids unneeded communication with the server for each subtree with explicit mergeinfo. If the merge target has many subtrees with mergeinfo, the performance improvment can be dramatic, see http://svn.haxx.se/dev/archive-2009-03/0198.shtml. Votes: +1: pburba, rhuijben, arfrever * r36149-r36151, r36156 Fix starting apache httpd in daemon mode from python versions with win32process support. (Found after upgrading my python to 2.6.1) Justification: Without this fix I can't run the testsuite as non-admin after installing 2.6.1. Notes: r36149 and r36150 contain the actual fix and r36151 and r36156 clean up the mess I left after performing only partial commits. Votes: +1: rhuijben, arfrever * r36509 Don't contact the server asking about implicit mergeinfo when the explicit mergeinfo tells us all we need to know about what to merge and what not to. Justification: Avoids needless communication with the server in some use cases. For users with lots of subtree mergeinfo and slow network connections the performance improvement should be quite significant. Votes: +1: pburba Approved changes: =================