Subversion SVN Changelog

Changelog for Subversion (48,243 revisions):

r869313: 2008-02-08 20:51:36 [pburba] Paul T. Burba

  /subversion/libsvn_client/merge.c  [diff, log, file]

 Follow-up to r29216, fix for issue #3094.
 
 * subversion/libsvn_client/merge.c
   (normalize_merge_sources): If we add any svn_location_segment_t's to the
   array initially populated by svn_client__repos_location_segments(), be sure
   the path members don't have leading '/'.
   

r869312: 2008-02-08 16:40:34 [pburba] Paul T. Burba

  [1.5.x] /subversion/tests/cmdline/merge_tests.py  [diff, log, file]

 Remove some cruft from r29237.
 
 * subversion/tests/cmdline/merge_tests.py
   (no_mergeinfo_from_no_op_merge): Remove a failure raise. Moral of the
   story: Don't do two things at once when one of those things is backporting.

r869311: 2008-02-08 15:34:16 [pburba] Paul T. Burba

  [1.5.x] /STATUS  [diff, log, file]
  [1.5.x] /subversion/libsvn_client/merge.c  [diff, log, file]
  [1.5.x] /subversion/tests/cmdline/merge_tests.py  [diff, log, file]

 Merge r29225 via svnmerge.py from trunk to 1.5.x.

r869310: 2008-02-08 15:32:38 [pburba] Paul T. Burba

  [1.5.x] /  [diff, log, file]
  [1.5.x] /STATUS  [diff, log, file]

 * STATUS: Vote for r29225, approving.

r869309: 2008-02-08 14:54:38 [kameshj] Kamesh Jayachandran

  [issue-2897] /subversion/libsvn_client/merge.c  [diff, log, file]

 On the issue-2897 branch:
 
 Improve documentation.
 
 * subversion/libsvn_client/merge.c
   (merge_cmd_baton_t.reflective_rev_affected_paths):
    Improve documentation.
 
 Suggested by: glasser

r869308: 2008-02-08 13:25:01 [kameshj] Kamesh Jayachandran

  /subversion/libsvn_client/merge.c  [diff, log, file]

 Make use of the sesspool (which is otherwise not used at all) for
 temporary ra_session.
 
 * subversion/libsvn_client/merge.c
   (get_full_mergeinfo): create temporary ra_session from a temporary pool.

r869307: 2008-02-08 04:45:32 [glasser] David Glasser

  /subversion/libsvn_ra_neon/commit.c  [diff, log, file]
  /subversion/libsvn_ra_neon/lock.c  [diff, log, file]
  /subversion/libsvn_ra_neon/session.c  [diff, log, file]

 Follow-up to r29231, making sure that every svn_ra_neon__request_t is
 explicitly destroyed even when errors are thrown. In two cases, this
 means adding blatantly missing svn_ra_neon__request_destroy calls; in
 the rest it just means changing SVN_ERR calls to "if (err) goto
 cleanup" and the like.
 
 (Full disclosure: I'm not going to be at the machine where I have
 davautocheck working until Monday, so this hasn't actually been fully
 tested yet.)
 
 * subversion/libsvn_ra_neon/commit.c
   (do_checkout, commit_delete_entry, commit_close_file): Always
    destroy the request.
 
 * subversion/libsvn_ra_neon/lock.c
   (do_lock, svn_ra_neon__get_lock): Always destroy the request.
 
 * subversion/libsvn_ra_neon/session.c
   (exchange_capabilities): Always destroy the request.

r869306: 2008-02-08 03:54:46 [kfogel] Karl Fogel

  /subversion/include/svn_wc.h  [diff, log, file]

 * subversion/include/svn_wc.h
   (svn_wc_conflict_choice_t): Fix comments, and reorder enum,
     following up to r29189.
 
 Found by: glasser
 (He noticed that the comments didn't match up.)

r869305: 2008-02-08 00:43:41 [glasser] David Glasser

  /subversion/libsvn_ra_neon/util.c  [diff, log, file]

 Fix the same problem fixed by r29230 from another angle. (Fortunately
 these two changes are two great tastes that taste great together.)
 
 In two cases where svn_ra_neon__request_destroy is called, make sure
 that it gets called even if errors occur. This means that we don't
 actually have to put any requirements on the pool passed to
 svn_ra_neon__parsed_request, as threatened in the log message for
 r29230.
 
 There's another handful of svn_ra_neon__request_create/destroy pairs
 scattered across the library that probably need this treatment as
 well.
 
 * subversion/libsvn_ra_neon/util.c
   (parsed_request): Add a request argument and move request creation
    and destruction to...
   (svn_ra_neon__parsed_request): ... here, so that the destruction
    always happens.
   (svn_ra_neon__simple_request): For good measure, make this
    function destroy its request even when errors are thrown, too.

r869304: 2008-02-08 00:13:39 [glasser] David Glasser

  /subversion/libsvn_ra_neon/get_location_segments.c  [diff, log, file]

 Fix a segfault on global pool cleanup after merge which has been
 occurring on some platforms since r28442. It appears that, due to the
 ne_decompress destruction in the request cleanup handler,
 svn_ra_neon__parsed_request must always be called with a pool whose
 lifetime is no longer than the RA session pool's. (This will be
 documented in a future revision. (Update: actually, r29231 fixes
 this in a different way.))
 
 * subversion/libsvn_ra_neon/get_location_segments.c
   (svn_ra_neon__get_location_segments): Run everything (including
    svn_ra_neon__parsed_request) in a subpool.

r869303: 2008-02-07 22:40:19 [epg] Eric Gillespie

  [1.5.x] /STATUS  [diff, log, file]

 Nominate r29106.

r869302: 2008-02-07 20:11:37 [kfogel] Karl Fogel

  /subversion/svn/conflict-callbacks.c  [diff, log, file]

 * subversion/svn/conflict-callbacks.c
   (svn_cl__conflict_handler): Use more human-friendly phrasing and
     formatting in the interactive conflict resolution prompt. The
     new longest prompt looks like this:
 
     "Select: (p)ostpone, all of (M)ine, all of (T)heirs, (r)esolved,
              (h)elp for more options:"
 
 Suggested by: cmpilato
               epg
 
 Unsolicited musical interlude by: glasser

r869301: 2008-02-07 18:57:20 [hwright] Hyrum Wright

  [1.5.x] /  [diff, log, file]
  [1.5.x] /STATUS  [diff, log, file]
  [1.5.x] /subversion/tests/cmdline/svntest/verify.py  [diff, log, file]

 Merge r29220 from trunk: Fix test framework issue that considers no actual
 output as OK, even if there's output expected.
 
 Approved by: +1: lgo, cmpilato

r869300: 2008-02-07 18:53:27 [cmpilato] C. Michael Pilato

  [1.5.x] /STATUS  [diff, log, file]

 Do some suggestin' and votin' and approvin'. Yee haw.

r869299: 2008-02-07 18:51:38 [cmpilato] C. Michael Pilato

  /subversion/libsvn_client/merge.c  [diff, log, file]
  /subversion/tests/cmdline/merge_tests.py  [diff, log, file]

 Make single-file merge notifications more like directory merge
 notifications in that header lines only precede blocks containing
 operative notifications.
 
 * subversion/libsvn_client/merge.c
   (IS_OPERATIVE_NOTIFICATION): New macro.
   (notification_receiver): Replace now-redundant conditional with use
     of the new IS_OPERATIVE_NOTIFICATION() macro.
   (single_file_merge_notify): Employ the IS_OPERATIVE_NOTIFICATION() macro.
 
 * subversion/tests/cmdline/merge_tests.py
   (no_mergeinfo_from_no_op_merge): Expect no output from no-op merges.

r869298: 2008-02-07 18:49:39 [lgo] Lieven Govaerts

  [1.5.x] /STATUS  [diff, log, file]

 * STATUS: Nominate r29220 for backport

r869297: 2008-02-07 15:17:23 [hwright] Hyrum Wright

  [1.5.x] /  [diff, log, file]
  [1.5.x] /STATUS  [diff, log, file]
  [1.5.x] /subversion/libsvn_client/log.c  [diff, log, file]

 Merge r29215 from trunk: Fix an RA session lifetime bug.
 
 Approved by: +1 cmpilato, hwright

r869296: 2008-02-07 15:13:42 [hwright] Hyrum Wright

  [1.5.x] /STATUS  [diff, log, file]

 * STATUS: Vote for r29215, approving.

r869295: 2008-02-07 11:23:10 [fabien] Fabien COELHO

  [1.5.x] /subversion/po/fr.po  [diff, log, file]

 French translation update for 1.5 branch (r29220).
 
 * subversion/po/fr.po: po-update + 2 fuzzy and 1 new

r869294: 2008-02-07 10:26:47 [lgo] Lieven Govaerts

  /subversion/tests/cmdline/svntest/verify.py  [diff, log, file]

 Fix issue in the test framework: a situation where svn doesn't return any output
 but we do expect some is now detected and considered an error.
 
 This breaks merge_test 62, but it looks like that test should be failing anyway.
 
 Found by: cmpilato
 
 * subversion/tests/cmdline/svntest/verify.py
   (ExpectedOutput.is_equivalent_list): if we expect output but there's none
    raise an error.

r869293: 2008-02-07 09:03:52 [cmpilato] C. Michael Pilato

  [1.5.x] /STATUS  [diff, log, file]

 Recommend r29218 for backport.

r869292: 2008-02-07 09:00:22 [cmpilato] C. Michael Pilato

  /subversion/libsvn_fs_base/fs.c  [diff, log, file]

 Fix some Ruby test failures introduced by r29076.
 
 * subversion/libsvn_fs_base/fs.c
   (base_open): Delay the writing of the format file until after the
     databases have been opened/created.

r869291: 2008-02-07 08:15:30 [cmpilato] C. Michael Pilato

  [1.5.x] /STATUS  [diff, log, file]

 Suggest some stuff for backport (and consistify some other stuff).

r869290: 2008-02-07 08:10:20 [cmpilato] C. Michael Pilato

  /subversion/libsvn_client/merge.c  [diff, log, file]
  /subversion/tests/cmdline/merge_tests.py  [diff, log, file]

 Finish issue #3094 - Merge source normalization falls over when source
 URL history doesn't extend back as far as request range.
 
 * subversion/libsvn_client/merge.c
   (normalize_merge_sources): Deal with two special cases: a) when the
     fetched source location segments don't fully cover the range of
     requested revisions, and b) when the first segment fetched is a gap
     segment. Were it not for my concern about performance when
     merging from pre-1.5 repositories, I would have take the much simpler
     approach noted in comments herein. What can I say? I'm a nice guy.
 
 * subversion/tests/cmdline/merge_tests.py
   (test_list): Remove XFail from merge_range_predates_history() test.

r869289: 2008-02-07 07:14:23 [cmpilato] C. Michael Pilato

  /subversion/libsvn_client/log.c  [diff, log, file]

 Fix an RA session lifetime issue.
 
 * subversion/libsvn_client/log.c
   (svn_client__get_copy_source): Use a subpool to manage the session lifetime.