- r984797: 2010-08-12 14:15:11 [neels] Neels J. Hofmeyr
-
/subversion/include/svn_client.h [diff, log, file]
* subversion/include/svn_client.h (svn_client_resolved): Tweak comment.
- r984781: 2010-08-12 13:56:25 [neels] Neels J. Hofmeyr
-
/subversion/libsvn_wc/update_editor.c [diff, log, file]
* subversion/libsvn_wc/update_editor.c (add_file):
Cosmetically rejigger auto-remove of tree conflicts against local
unversioned obstructions. Also possibly fix some future case where a node
has other conflicts beside a tree conflict (currently not possible AFAIK),
by checking for conflicts again after a tree conflict has been
automatically removed.
- r984758: 2010-08-12 13:18:08 [neels] Neels J. Hofmeyr
-
/subversion/libsvn_wc/wc-metadata.sql [diff, log, file]
* subversion/libsvn_wc/wc-metadata.sql (WORKING_NODE):
Tweak comment on the 'keep_local' column.
- r984751: 2010-08-12 12:58:18 [neels] Neels J. Hofmeyr
-
/subversion/libsvn_wc/update_editor.c [diff, log, file]
* subversion/libsvn_wc/update_editor.c (add_file): Cosmetic.
ERR is usually the last variable declared. Have separate lines for
two other var declarations.
- r982056: 2010-08-03 22:45:42 [neels] Neels J. Hofmeyr
-
/subversion/libsvn_client/merge.c [diff, log, file]
* subversion/libsvn_client/merge.c (merge_file_changed):
Treat svn_depth_unknown as svn_depth_infinity when checking whether to
skip child nodes. Because of r981470, this doesn't fix anything anymore,
but it still seems the right thing to do.
- r982042: 2010-08-03 21:46:09 [neels] Neels J. Hofmeyr
-
/subversion/libsvn_wc/update_editor.c [diff, log, file]
* subversion/libsvn_wc/update_editor.c (close_directory):
Use SVN_WC__SINGLE_DB instead of SINGLE_DB (chosen wrongly in r981470).
Suggested by: rhuijben
- r981470: 2010-08-02 12:05:27 [neels] Neels J. Hofmeyr
-
/subversion/libsvn_wc/update_editor.c [diff, log, file]
SINGLE_DB: Don't write depth = "unknown" for dirs added on update.
* subversion/libsvn_wc/update_editor.c
(close_directory):
Translate depth_unknown to _infinity on "second pass" of adding a dir.
- r966569: 2010-07-22 10:42:17 [neels] Neels J. Hofmeyr
-
/subversion/libsvn_wc/wc-metadata.sql [diff, log, file]
* subversion/libsvn_wc/wc-metadata.sql (NODE_DATA): Tweak comment.
- r966048: 2010-07-20 23:05:21 [neels] Neels J. Hofmeyr
-
/subversion/tests/cmdline/commit_tests.py [diff, log, file]
* subversion/tests/cmdline/commit_tests.py (commit_url): Fix test failure.
- r965912: 2010-07-20 17:08:14 [neels] Neels J. Hofmeyr
-
/notes/tree-conflicts/all-add-vs-add-tree-conflicts.txt [diff, log, file]
/subversion/libsvn_wc/update_editor.c [diff, log, file]
/subversion/tests/cmdline/checkout_tests.py [diff, log, file]
/subversion/tests/cmdline/schedule_tests.py [diff, log, file]
/subversion/tests/cmdline/svntest/actions.py [diff, log, file]
/subversion/tests/cmdline/switch_tests.py [diff, log, file]
/subversion/tests/cmdline/update_tests.py [diff, log, file]
Don't interrupt the update/switch/checkout with an error on incoming dir add
vs. local unversioned. Instead, flag a conflict (so far a tree conflict) and
pull in all the other unrelated nodes.
Where r959735 dealt with add_file(), this deals with add_directory(). Until
single-db, incoming dir can only improve vs. local unversioned. ...Also
because add_directory() doesn't get copy_from info like add_file().
Note, there is bound to be some discussion on whether to flag a conflict on
local unversioned. Obviously, I'm +1 for flagging a conflict to make the path
more visible, but instead we could also just notify. At least don't bail.
* subversion/libsvn_wc/update_editor.c
(add_directory):
Don't bail on local unversioned, instead flag a tree conflict. Also remove
some obsolete logic and streamline. Change the flow of dirs with local
status == _obstructed: set add_existed = TRUE, but explicitly notify with
'A' like the old client would (also see comment in this patch starting
"Pre-single-db, ..." and schedule_tests:delete_redelete_fudgery).
* subversion/tests/cmdline/checkout_tests.py
(checkout_with_obstructions,
forced_checkout_of_dir_with_file_obstructions),
* subversion/tests/cmdline/switch_tests.py
(bad_intermediate_urls,
forced_switch_failures,
switch_with_obstructing_local_adds):
* subversion/tests/cmdline/update_tests.py
(obstructed_update_alters_wc_props,
forced_update_failures):
Change these tests to match the new tree conflicts behavior. Some of the
tests are partly rewritten from scratch (using svntest.factory).
* subversion/tests/cmdline/schedule_tests.py
(test_list):
Remove XFail from delete_redelete_fudgery(), as the way 'obstructed' dirs
are treated now fixes that error (see also above).
* subversion/tests/cmdline/svntest/actions.py
(run_and_verify_checkout2): New function to have an explicit argument saying
whether the checkout target WC should be removed or not, instead of
determining that by presence of a --force argument, which is horrible.
(run_and_verify_checkout): Move guts to run_and_verify_checkout2() and call
that so that this function behaves as it always did.
* notes/tree-conflicts/all-add-vs-add-tree-conflicts.txt: Update.
- r965877: 2010-07-20 15:23:43 [neels] Neels J. Hofmeyr
-
/subversion/tests/cmdline/svntest/factory.py [diff, log, file]
* subversion/tests/cmdline/svntest/factory.py: Add an error check.
- r959983: 2010-07-02 13:43:27 [neels] Neels J. Hofmeyr
-
/subversion/libsvn_client/info.c [diff, log, file]
* subversion/libsvn_client/info.c (info_found_node_callback):
This is mostly indents. Simplify some ghastly if-clause structuring.
- r959738: 2010-07-01 18:21:57 [neels] Neels J. Hofmeyr
-
/subversion/tests/cmdline/switch_tests.py [diff, log, file]
Follow-up on r959735.
* subversion/tests/cmdline/switch_tests.py
(forced_switch_failures):
Remove a unix-EOL that caused an expected-output mismatch on windows.
- r959735: 2010-07-01 17:51:36 [neels] Neels J. Hofmeyr
-
/notes/tree-conflicts/all-add-vs-add-tree-conflicts.txt [log, file]
/subversion/libsvn_client/info.c [diff, log, file]
/subversion/libsvn_wc/update_editor.c [diff, log, file]
/subversion/tests/cmdline/checkout_tests.py [diff, log, file]
/subversion/tests/cmdline/switch_tests.py [diff, log, file]
/subversion/tests/cmdline/tree_conflict_tests.py [diff, log, file]
/subversion/tests/cmdline/update_tests.py [diff, log, file]
Improve selected add-vs.-add tree conflict situations -- add as in unversioned
obstruction, add, copy-here, move-here, upon update, switch and checkout.
Take advantage of the wc-ng we already have to teach update_editor:add_file()
and close_file() to treat many add-vs-add conflicts much better. Note that
this is not changing *_directory() yet.
A very detailed description of this patch can be found in
notes/tree-conflicts/all-add-vs-add-tree-conflicts.txt
, which is being added in this revision.
* subversion/libsvn_wc/update_editor.c
(complete_directory):
Don't remove not_present nodes that were created to mark an add-vs.-add
conflict against unversioned.
(struct file_baton):
New flag ADDING_BASE_UNDER_LOCAL_ADD.
(create_tree_conflict): New function.
(check_tree_conflict):
Out-source the conflict struct creation to create_tree_conflict() above.
Fix the comment.
(build_write_revert_base_props): New function.
(add_file):
Remove some obsolete logic, but, in essence, just streamline and flag a
few more tree conflicts. Also use the COPYFROM_* args to flag conflicts
where local add hits an incoming copy, and to omit conflicts where local
and incoming are identical copies.
(close_file):
Detect whether incoming adds are symlinks, and flag tree conflicts if
symlink-ness mismatches local. Query file_baton's new parameter
ADDING_BASE_UNDER_LOCAL_ADD, and pull the incoming add directly into the
revert-base slots, if appropriate, not changing the WORKING/ACTUAL data.
* subversion/libsvn_client/info.c
(info_found_node_callback):
Make sure that unversioned nodes can show a tree conflict on 'svn info'.
* notes/tree-conflicts/all-add-vs-add-tree-conflicts.txt:
New file.
* subversion/tests/cmdline/checkout_tests.py
(co_with_obstructing_local_adds),
* subversion/tests/cmdline/svntest/actions.py
(run_and_validate_lock),
* subversion/tests/cmdline/update_tests.py
(forced_update_failures, update_with_obstructing_additions):
* subversion/tests/cmdline/switch_tests.py
(obstructed_switch, forced_switch_failures,
switch_with_obstructing_local_adds):
Change these tests to match the new tree conflicts behavior. Some of the
tests are rewritten from scratch (using svntest.factory).
* subversion/tests/cmdline/tree_conflict_tests.py
(incoming_paths, localmod_paths, modify, set_up_repos, ensure_tree_conflict):
Add 'F3' and a flag to the test suite so that the incoming copy always
mismatches the local copy (for files only, so far).
- r959678: 2010-07-01 14:29:10 [neels] Neels J. Hofmeyr
-
/subversion/tests/manual/permutations.py [old log]
/subversion/tests/manual/tree-conflicts-add-vs-add.py [diff, log, file]
Tweak tree conflicts add-vs-add manual tests.
* subversion/tests/manual/permutations.py:
Remove file, has always been obsolete because of itertools.product().
(Suggested by danielsh, thanks)
* subversion/tests/manual/tree-conflicts-add-vs-add.py:
Use itertools.product instead of permutations.py. Also modify the
copy-source in copy cases, so that I can see whether local mods are
carried over from the copy source's WC state.
- r959383: 2010-06-30 17:38:05 [neels] Neels J. Hofmeyr
-
/subversion/tests/manual/tree-conflicts-add-vs-add.py [diff, log, file]
* subversion/tests/manual/tree-conflicts-add-vs-add.py: Add tests for checkout.
- r959139: 2010-06-29 23:13:21 [neels] Neels J. Hofmeyr
-
/subversion/libsvn_wc/node.c [diff, log, file]
Fix r959112 for good, hopefully.
* subversion/libsvn_wc/node.c
(svn_wc__node_get_copyfrom_info): Don't forget to initialize out params.
- r959119: 2010-06-29 21:51:06 [neels] Neels J. Hofmeyr
-
/subversion/bindings/javahl/native/CreateJ.cpp [diff, log, file]
Another fix for r959112: Try to un-break the javahl bindings.
* subversion/bindings/javahl/native/CreateJ.cpp
(CreateJ::Status):
Pass NULL for the two new parameters of svn_wc__node_get_copyfrom_info().
- r959115: 2010-06-29 21:46:40 [neels] Neels J. Hofmeyr
-
/subversion/libsvn_wc/node.c [diff, log, file]
Fix something in r959112.
* subversion/libsvn_wc/node.c
(svn_wc__node_get_copyfrom_info):
Use relpath_join instead of dirent_join in one place.
Suggested by: Bert
- r959112: 2010-06-29 21:36:28 [neels] Neels J. Hofmeyr
-
/subversion/include/private/svn_wc_private.h [diff, log, file]
/subversion/libsvn_client/commit_util.c [diff, log, file]
/subversion/libsvn_client/copy.c [diff, log, file]
/subversion/libsvn_client/export.c [diff, log, file]
/subversion/libsvn_client/info.c [diff, log, file]
/subversion/libsvn_client/ra.c [diff, log, file]
/subversion/libsvn_client/url.c [diff, log, file]
/subversion/libsvn_wc/node.c [diff, log, file]
Teach svn_wc__node_get_copyfrom_info() to also return the URL as separate root
and relpath. There should be no "real" changes with this patch except some
slight micro-optimisations (this isn't used yet, breaking up a larger patch).
* subversion/include/private/svn_wc_private.h,
* subversion/libsvn_wc/node.c
(svn_wc__node_get_copyfrom_info):
Add parameters COPYFROM_ROOT_URL and COPYFROM_REPOS_RELPATH.
* subversion/libsvn_client/commit_util.c (harvest_committables),
* subversion/libsvn_client/copy.c (calculate_target_mergeinfo, try_copy),
* subversion/libsvn_client/export.c (copy_versioned_files),
* subversion/libsvn_client/info.c (build_info_for_entry),
* subversion/libsvn_client/ra.c (svn_client__repos_locations),
* subversion/libsvn_client/url.c (svn_client__entry_location):
Pass NULL for the two new parameters of svn_wc__node_get_copyfrom_info(),
except, when only the presence of copyfrom information is tested, don't
bother to get a full URL and ask for the COPYFROM_REPOS_ROOT only.
- r959091: 2010-06-29 20:27:38 [neels] Neels J. Hofmeyr
-
/subversion/tests/cmdline/svntest/factory.py [diff, log, file]
* subversion/tests/cmdline/svntest/factory.py: Minor improvements.
- r959090: 2010-06-29 20:22:19 [neels] Neels J. Hofmeyr
-
/subversion/tests/manual [log, file]
/subversion/tests/manual/permutations.py [log, file]
/subversion/tests/manual/README [log, file]
/subversion/tests/manual/tree-conflicts-add-vs-add.py [log, file]
Add a manual test script for add-vs-add tree conflicts.
* subversion/tests/manual: New directory.
* subversion/tests/manual/README,
subversion/tests/manual/tree-conflicts-add-vs-add.py,
subversion/tests/manual/permutations.py: New files.
- r941975: 2010-05-07 00:05:48 [neels] Neels J. Hofmeyr
-
/subversion/tests/cmdline/svntest/factory.py [diff, log, file]
* subversion/tests/cmdline/svntest/factory.py (TestFactory):
Make a distiction between quiet and unquiet status.
- r941974: 2010-05-07 00:02:59 [neels] Neels J. Hofmeyr
-
/subversion/include/svn_wc.h [diff, log, file]
/subversion/libsvn_subr/io.c [diff, log, file]
/subversion/libsvn_wc/adm_ops.c [diff, log, file]
/subversion/libsvn_wc/node.c [diff, log, file]
/subversion/libsvn_wc/wc_db.h [diff, log, file]
/subversion/tests/cmdline/copy_tests.py [diff, log, file]
/subversion/tests/cmdline/svntest/wc.py [diff, log, file]
/subversion/tests/cmdline/tree_conflict_tests.py [diff, log, file]
Some cosmetic changes.
* subversion/include/svn_wc.h
(svn_wc_remove_from_revision_control): typo.
* subversion/libsvn_subr/io.c
(svn_io_sleep_for_timestamps): typo.
* subversion/libsvn_wc/adm_ops.c
(svn_wc_delete4): line wrap.
* subversion/libsvn_wc/node.c
(svn_wc__node_get_copyfrom_info): blank line.
* subversion/libsvn_wc/wc_db.h
(svn_wc__db_op_read_tree_conflict): typo.
* subversion/tests/cmdline/copy_tests.py
(mixed_wc_to_url): Add two failure messages.
* subversion/tests/cmdline/svntest/wc.py: typo.
* subversion/tests/cmdline/tree_conflict_tests.py
(force_del_tc_inside): Compress two tweak()s into one.
- r941472: 2010-05-05 20:28:30 [neels] Neels J. Hofmeyr
-
/subversion/libsvn_client/commit_util.c [diff, log, file]
Fix ill var naming from r941094.
* subversion/libsvn_client/commit_util.c
(harvest_committables):
Rename sub-scoped copyfrom_url/_rev to node_*, to get rid of a compiler
warning.