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 1.5.0:
============================

Candidate changes for 1.5.1:
============================

 * r31313
   Make all ra layers return a consistent error when a path is not found in the
   repository.
   Notes:
     Small change, but API consistency is always a good thing.
     Does not merge cleanly!
   Votes:
     +1: lgo, cmpilato
     -0: markphip (serf tests fail on OSX), epg (failure confirmed)

 * r31196, r31199, r31280, r31283, r31287
   Don't commit an add of a missing item (issue #3198).
   Notes:
     r31199 is a trivial followup that adds a missing SVN_ERR() wrap.
     r31280, r31283 and r31287 add a regression test.
   Votes:
     +1: kfogel (r31196 and r31199 only)
     +0: danielsh
     +1: lgo

 * r31228
   Fix issue #2721: don't croak on invalid dates in 'log' and 'blame'.
   Justification:
     Let the commands be useful even when some revisions' dates are corrupt.
   Votes:
     +1: danielsh, epg

 * r30748
   Don't try to drive the merge editor when there is nothing to be merged.
   Justification:
     I didn't originally nominate this for backport as I couldn't
     demonstrate any problem with it on trunk.  But the new merge test in
     the r31059 group ("don't explicitly record implicit mergeinfo") does fail
     without it and thus neatly demonstrates why this change is needed.
   Votes:
     +1: pburba, markphip

 * r31059, 31060, 31061, 31075, 31151
   Fix issue #3157, Merging a change from a path's natural history
   creates self-referential mergeinfo
   Notes:
     r31075 and 31151 are the core fixes for the issue, the other revisions
     are a new test for it (and a few minor follow-ups to that test).
     The addition of the new merge test in r31059 trivally conflicts
     because several other new merge tests are not backported yet.
     The new test fails if r30748 isn't backported first.
   Votes:
     +1: cmpilato, markphip
     -1: pburba (looks like there are problems with subtrees, vetoing
         until I can look into these further).

 * r31300, r31301, r31312
   Fix issue #3199, subtree merges broken when required ranges don't
   intersect with merge target.
   Justification:
     A bit of an edge case, but when it is encountered merge silently
     fails to do the correct merge but sets mergeinfo as if it did.
   Notes:
     r31300 is a test for this issue, r31301 fixes the core problem where
     the part of the merge was skipped, and r31312 fixes a problem with
     merge notification.
   Votes:
     +1: pburba

 * r31425
   Support the 'http-library' option when Subversion was configured
   with the '--enable-runtime-module-search' option.
   Votes:
     +1: arfrever

 * r30915
   Fix issue #2887. 'svn status --xml' was generating invalid XML
   when reporting the status of a file which resides in an untracked
   directory.
   Notes:
     Fixed a while back in trunk. The problem was brought up again
     later by a user on the dev list:
     http://subversion.tigris.org/servlets/ReadMsg?list=dev&msgNo=139158
   Votes:
     +1: stsp, danielsh

 * r31504
   Optimize file merges performed with ancestrally related sources by
   using the svn_ra_get_log2() interface to figure out in which revisions
   the source really changed, and using that information to filter out
   no-op revision ranges prior to merging.
   Votes:
     +1: cmpilato

 * r31619
   ra_serf: Fixes an endless loop when a user provides incorrect credentials
   3 times and a segfault when a user cancels the TSVN credentials prompt.
   Votes:
     +1: lgo

 * r31692, r31695
   Fix issue #2939 over ra_neon/serf:  bogus existence check.
   Votes:
     +1: epg

Approved changes (all releases):
================================

Approved changes (1.5.1 and later):
===================================

 * (r31159, r31179) or r31193 from 1.5.x-r31159-r31179 branch.
   Fix "file not found" error when a merge target is a broken symbolic link.
   Notes:
     r31179 just makes sure the new test is skipped on Windows.
     r31193 from 1.5.x-r31159-r31179 branch has r31159 and r31179 
     merged to it with a conflict resolution.
   Notes:
     This was merged to branch in r31218 and later reverse-merged so that
     it will be a 1.5.1 candidate and not part of 1.5.0
   Votes:
     +1: kfogel(r31159, r31179), kameshj, cmpilato (r31159, r31179)

 * r30746, r30747
   Fix a minor merge range notification header bug.
   Notes:
     r30747 is just cleanup of some cruft that snuck in with r30746,
     thus adhering to my "every commit should need a follow-up" ethos.
     r30746 contains, like, one tiny *real* change, then a whole bunch
     of bogus indentation changes that were supposed to be part of
     r30748.  Probably best to not fix that indentation during
     backport, in case we later backport r30748.
   Notes:
     This was merged to branch in r31186 and later reverse-merged so that
     it will be a 1.5.1 candidate and not part of 1.5.0
   Votes:
     +1: pburba, markphip, cmpilato

 * r30949, r30962
   Fix issue #3188.  Mergeinfo on switched targets/subtrees should elide
   to repos.
   Notes:
     This was merged to branch in r31185 and later reverse-merged so that
     it will be a 1.5.1 candidate and not part of 1.5.0
   Votes:
     +1: pburba, markphip, cmpilato

 * r31603
   Export svn_path_is_url to the bindings.
   Votes:
     +1: epg, blair

 * r31223
   Fix segfault when svn_ra_open3 is passed a bogus URL such as 'bogusURL'.
   Notes:
     Not possible to make svn(1) segfault with this, so can wait for 1.5.1.
   Votes:
     +1: epg, stsp, danielsh