* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * THIS RELEASE STREAM IS OPEN TO BUG FIXES. * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * This file tracks the status of releases in the 1.5.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. Candidate changes for 1.5.8: ============================ * r36106 Resolves issue #3374, "svn cp --parents path URL encoding issues" For svn cp --parent ^/fö ^/dö/fö The --parents code checked and might creat^/d%C3%B6, while the file was copied to ^/dö as expected. Justification: Without this patch --parents is broken for utf8 characters in repos to repos copy and move operations. Votes: +1: rhuijben -0: gstein (encoding issues are tricky; I'd rather tell people to upgrade to 1.6.0, rather than 1.5.6) * r36111 On Windows don't retry deleting a file 100 times when this file is really not a file but just a directory. Justification: Makes removing unversioned directories much faster. (You can see the delay with: svn rm --force ) Votes: +1: rhuijben -0: gstein (edge case performance fixes shouldn't be backported; if the user wants the fix, they can upgrade to 1.6.0 rather than 1.5.6) * r36527 Yet another merge performance improvement: Potentially reduce server queries about subtree mergeinfo during reverse merges. Justification: Large numbers of subtrees with mergeinfo can significantly slow merge performance, this alleviates that somewhat in a common use case. Votes: +1: pburba * r36730, r36766, r36775 Fix issue #3383 'Foreign repository merge can add mergeinfo from the foreign repos' Justification: At best this mergeinfo refers to non-existent paths in the target's repos. At worst the mergeinfo intersects with actual path/revisions in the merge target. The latter might cause real problems, making it appear merges have been done that actually have not or vice-versa. Notes: r36775 is the fix, the other revisions are expansion of an existing merge test to cover this issue. There are some minor conflicts, a backport branch can be found at http://svn.collab.net/repos/svn/branches/1.5.x-r36775. Votes: +1: pburba, arfrever * r37072, r37083 Fix issue #3392 'Parsing error with reverse merges and non-inheritable mergeinfo' Justification: Can cause repeat merges to break in the presence of non-inheritable mergeinfo. Notes: r37072 is a test for the issue, r37083 is the fix. There is a backport branch at http://svn.collab.net/repos/svn/branches/1.5.x-issue3392, but it is dependent on r36879 being merged first. (If this change *must* be merged before r36879 I'll create a branch to do so - pburba) Votes: +1: pburba, arfrever * r37137 Make some sweeping pool lifetime fixes in the BDB backend. Justification: Bringing 'svn ls -v' over HTTP on a directory with 10000 files from 2Gb+ of RAM to under 400Mb is a very good thing. Notes The merge conflicts in a few places due to some minor cleanups performed post-1.5.x. A backport branch resolving it can be found here: http://svn.collab.net/repos/svn/branches/1.5.x-r37137 Votes: +1: cmpilato, arfrever * r37768 Make svnadmin load tolerate mergeinfo with "\r\n". Justification: Prior to this change you cannot load a dump file that contains Windows style newlines in svn:mergeinfo. See http://svn.haxx.se/dev/archive-2009-05/0179.shtml. Votes: +1: pburba, cmpilato * r37779 Fix a potential segfault when doing noop file merges. Justification: Not common, but a segfault is a segfault. See http://svn.haxx.se/dev/archive-2009-05/0363.shtml. Notes: There are text conflicts in both merge.c and merge_tests.py, a backport branch resolving these can be found at: http://svn.collab.net/repos/svn/branches/1.5.x-r37779. Votes: +1: pburba * r38068 Fix issue #3119, "File '...' already exists" when it obviously doesn't, during commit. Justification: This was introduced in 1.5.1 and propagated till now. Votes: +1: stylesen * r38686-7 Fix serious bugs in libsvn_ra_svn and svnsync. (If svnsync errors while syncing from svnserve to mod_dav_svn, it commits a half-written transaction rather than aborting it.) Justification: A potentially large audience of users may unknowingly be creating corrupt repository mirrors! Notes: Running 'svn merge -r36685:36687 ^/truenk' creates a spurious one-line conflict for some reason, perhaps due to linefeeds. Easy to clean up. Votes: +1: sussman * r39019 Fix bug where svn_[rangelist|mergeinfo]_[merge|intersect|remove|diff] APIs can modify their *non*-output arguments. Justification: No reports of this causing any problems that I know of, which is probably due to the fact that users of an API like svn_mergeinfo_merge typically only care about the output arguments. The new C tests added to mergeinfo-test.c clearly demonstrate the bug. Branch: ^/branches/1.5.x-r39109 Votes: +1: pburba * r36023 Resolve symlinks when checking ~/.subversion Justification: It gives flexibility to users to link different .subversion folders. Votes: +1: stylesen, stsp * r39219 Perform MIME type matching case-insensitively. (Issue #3479) Justification: The broken behavior is inconsistent with the way MIME mapping files are used everywhere else in the world. Votes: +1: cmpilato * r40008 Respect Apache's ServerSignature directive. Justification: Many subversion server administrators want it. Votes: +1: stylesen Approved changes: ================= * r36851, r36857 'svnadmin upgrade'-ing pre-1.2 repositories fails for lack of 'db/format' file. Justification: 'svnadmin upgrade' was designed to allow for this situation, but clearly does not. Notes: Backport branch needed, the function in question has seen a lot of churn. Votes: +1: cmpilato, danielsh, rhuijben -0: hwright (until such time as a backport branch exists)