STATUS

This file tracks the status of releases in the 1.4.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.

NOTE:  This branch uses svnmerge.py for merge tracking.  Please familiarize
yourself with that tool if you plan to actually merge changes here.

For example, to merge a number of revisions from trunk into branch,
you can use the svnmerge.py script that is in the branch.

$ contrib/client-side/svnmerge.py merge -r 21000-21006,21008,21876
$ svn diff
$ svn commit

Do not run svnmerge.py N separate times for each revision to merge
over, since svnmerge.py will not merge into a WC with uncommitted
changes.

Status of 1.4.6:
================

Candidate changes:

  * r22355
    Put repository and WC uuids in the correct order in a "svn switch
    --relocate" error message.
    Justification:
      Confusing error messages are Bad.
    Votes:
      +1: glasser, hwright

  * r21989, r22884, r23347-23349, r23357, r23359, r23363, r23364, r23384,
    r23385, r23387, r23395, r23763, r23764, r23794, r23795, r23799, r23807
    Bring mucc and mucc-test.py up to date with respect to trunk,
    except for r23427, which uses svn_stringbuf_from_file2() in 1.5.
    Justification:
      This is becoming a very useful tool for making restructuring
      commits that would otherwise require extremely large working
      copies.
    Notes:
      Blair ran mucc-test.py against this merge and it passed the
      test.  The person doing the merge should ensure that mucc-test.sh is
      deleted, as svnmerge.py skips that for some reason.  Also, if
      a test merge run is done with without committing and then
      reverting, then also delete the new mucc-test.py, since the
      next svnmerge.py will ignore it the second time around.
    Votes:
      +1: blair, cmpilato
      -0: glasser (while backporting mucc is a reasonable idea, it has changed 
           a lot since this was proposed, including changing its name; either
           don't backport at all or backport even more)

  * r25178, -461
    Fix issue 2712: If 'anon-access' is 'read' but anonymous access is
    disallowed via authz rules, updates/checkouts won't work, even for users
    with appropriate permissions.
    Justification:
      User reported problem, small fix, includes test.
    Branch: 1.4.x-r25178
    Votes:
      +1: vgeorgescu

  * r25520, r25595
    Fix sporadic file and directory removal failing on Windows.
    Justification:
      On some systems, the problem is severe enough that 'svnadmin load'
      cannot complete on even relatively small repositories.
    Notes:
      There's a small merge conflict when merging because the directory rewind
      code is different on trunk, but the branch's rewind code is already
      limited to OSX, so the trunk code to prevent it on Windows is not
      needed.
    Votes:
      +1: djh

  * r26077
    Alow Subversion to be built against Neon 0.26.4 (by default).
    Justification:
      There's no reason to complicate building by requiring that
      --disable-neon-version-check be passed to configure.
    Votes:
      +1: dlr, hwright

Veto-blocked changes:

  * r21531, r21848, r21934
    Allow HTTP auth protocols to be configurable by way of a
    "http-auth-types" parameter.
    Justification:
      Allow GSSAPI/SSPI auth to be enabled/disabled on Windows with
      Neon 0.26.1+, addressesing numerous bug reports from
      TortoiseSVN, Subversion, Subclipse, etc.
    Votes:
      +1: dlr, djh, jerenkrantz
      -1: cmpilato (new features showing up in patch releases?  nuh-uh.)

  * r19808-19810, r22853
    Synchronise license text and then bump copyright dates to 2007.
    Justification:
      It's 2007.
    Notes: We might want to wait until we actually merge something we
      originally committed in 2007, possibly hold until 1.4.4?
    Votes:
      +1: malcolm, dlr, hwright
      -1: jerenkrantz (This is an unnecessary change to backport.)

 * r24497, r24504
   Make Perl-based txdelta_window_handler actually usable by editors
   and related functions.
   Justification:
     Binding completeness.
   Votes:
     +1: clkao
     +0: glasser (I don't grok SWIG, but having this available in Perl is good)
     -1: djames (see http://svn.haxx.se/dev/archive-2007-05/0058.shtml)

Approved changes:

  * r25833
    Fix administrative property file handling for prop changes to a
    file which is subsequently copied/moved over another
    schedule-delete file, where the original doesn't have any props
    (e.g. 'svn mv a b; svn ps name val b; svn mv b a' is a
    1.5-specific way to trigger this problem).
    Justification:
      WC library bug fix.
    Branch: TBD
    Notes:
      glasser tried to merge this but it really conflicts.  Can one of
      the people who voted for it merge it?
    Votes:
      +1: dionisos, dlr, hwright