You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Ed Price <ed...@gmail.com> on 2006/10/11 18:52:00 UTC
[PATCH] fix for revert notification bug (1.4.0 regression)
On 10/11/06, Ed Price <ed...@gmail.com> wrote:
> 1. Please see thread here:
>
> http://svn.haxx.se/users/archive-2006-10/0540.shtml
>
> I expect my patch broke it, so I'll try to fix it! :)
> Please confirm that I may file an issue.
Please find the fix (including new testcases) attached.
I suggest applying to trunk and nominating for 1.4.x backport,
as it does fix a regression in 1.4.0 and seems quite safe.
I'm happy to rev the patch if there are any issues with it.
Please let me know!
Thanks,
-Ed
[[[
Notify properly on property-only revert. Fixes a bug introduced
by the fix for issue #2517 (r18842).
* subversion/libsvn_wc/adm_ops.c
(revert_admin_things): Set '*reverted' when reinstalling
properties.
* subversion/tests/cmdline/revert_tests.py
(revert_propset__dir): New test.
(revert_propset__file): New test.
(revert_propdel__dir): New test.
(revert_propdel__file): New test.
(test_list): Add new tests.
Patch by: Ed Price <ed...@gmail.com>
]]]
Re: [PATCH] fix for revert notification bug (1.4.0 regression)
Posted by "Hyrum K. Wright" <hy...@mail.utexas.edu>.
Ed Price wrote:
> On 10/11/06, Ed Price <ed...@gmail.com> wrote:
>> 1. Please see thread here:
>>
>> http://svn.haxx.se/users/archive-2006-10/0540.shtml
>>
>> I expect my patch broke it, so I'll try to fix it! :)
>> Please confirm that I may file an issue.
>
> Please find the fix (including new testcases) attached.
>
> I suggest applying to trunk and nominating for 1.4.x backport,
> as it does fix a regression in 1.4.0 and seems quite safe.
>
> I'm happy to rev the patch if there are any issues with it.
> Please let me know!
Ping...
Has any of the committers had a chance to take a look at Ed's patch? If
nothing happens in the next few days, I'll file an issue.
-Hyrum
> [[[
>
> Notify properly on property-only revert. Fixes a bug introduced
> by the fix for issue #2517 (r18842).
>
> * subversion/libsvn_wc/adm_ops.c
> (revert_admin_things): Set '*reverted' when reinstalling
> properties.
>
> * subversion/tests/cmdline/revert_tests.py
> (revert_propset__dir): New test.
> (revert_propset__file): New test.
> (revert_propdel__dir): New test.
> (revert_propdel__file): New test.
> (test_list): Add new tests.
>
> Patch by: Ed Price <ed...@gmail.com>
>
> ]]]
>
>
> ------------------------------------------------------------------------
>
> Index: subversion/libsvn_wc/adm_ops.c
> ===================================================================
> --- subversion/libsvn_wc/adm_ops.c (revision 21887)
> +++ subversion/libsvn_wc/adm_ops.c (working copy)
> @@ -1702,10 +1702,13 @@
> /* Reinstall props if we need to. Only rewrite the baseprops,
> if we're reverting a replacement. This is just an optimization. */
> if (baseprops)
> - SVN_ERR(svn_wc__install_props(&log_accum, adm_access, name, baseprops,
> - baseprops,
> - entry->schedule == svn_wc_schedule_replace,
> - pool));
> + {
> + SVN_ERR(svn_wc__install_props(&log_accum, adm_access, name, baseprops,
> + baseprops,
> + entry->schedule == svn_wc_schedule_replace,
> + pool));
> + *reverted = TRUE;
> + }
>
> /* Clean up the copied state if this is a replacement. */
> if (entry->schedule == svn_wc_schedule_replace
> Index: subversion/tests/cmdline/revert_tests.py
> ===================================================================
> --- subversion/tests/cmdline/revert_tests.py (revision 21887)
> +++ subversion/tests/cmdline/revert_tests.py (working copy)
> @@ -583,7 +583,55 @@
> svntest.actions.run_and_verify_svn(None, [], [], "diff", wc_dir_2)
> svntest.actions.run_and_verify_svn(None, [], [], "revert", "-R", wc_dir_2)
>
> +def revert_propset__dir(sbox):
> + "revert a simple propset on a dir"
>
> + sbox.build()
> + wc_dir = sbox.wc_dir
> + a_path = os.path.join(wc_dir, 'A')
> + svntest.main.run_svn(None, 'propset', 'foo', 'x', a_path)
> + expected_output = "Reverted '" + a_path + "'"
> + svntest.actions.run_and_verify_svn(None, expected_output, [], "revert",
> + a_path)
> +
> +def revert_propset__file(sbox):
> + "revert a simple propset on a file"
> +
> + sbox.build()
> + wc_dir = sbox.wc_dir
> + iota_path = os.path.join(wc_dir, 'iota')
> + svntest.main.run_svn(None, 'propset', 'foo', 'x', iota_path)
> + expected_output = "Reverted '" + iota_path + "'"
> + svntest.actions.run_and_verify_svn(None, expected_output, [], "revert",
> + iota_path)
> +
> +def revert_propdel__dir(sbox):
> + "revert a simple propdel on a dir"
> +
> + sbox.build()
> + wc_dir = sbox.wc_dir
> + a_path = os.path.join(wc_dir, 'A')
> + svntest.main.run_svn(None, 'propset', 'foo', 'x', a_path)
> + svntest.main.run_svn(None, 'commit', '-m', 'ps', a_path)
> + svntest.main.run_svn(None, 'propdel', 'foo', a_path)
> + expected_output = "Reverted '" + a_path + "'"
> + svntest.actions.run_and_verify_svn(None, expected_output, [], "revert",
> + a_path)
> +
> +def revert_propdel__file(sbox):
> + "revert a simple propdel on a file"
> +
> + sbox.build()
> + wc_dir = sbox.wc_dir
> + iota_path = os.path.join(wc_dir, 'iota')
> + svntest.main.run_svn(None, 'propset', 'foo', 'x', iota_path)
> + svntest.main.run_svn(None, 'commit', '-m', 'ps', iota_path)
> + svntest.main.run_svn(None, 'propdel', 'foo', iota_path)
> + expected_output = "Reverted '" + iota_path + "'"
> + svntest.actions.run_and_verify_svn(None, expected_output, [], "revert",
> + iota_path)
> +
> +
> ########################################################################
> # Run the tests
>
> @@ -600,6 +648,10 @@
> revert_after_second_replace,
> revert_after_manual_conflict_resolution__text,
> revert_after_manual_conflict_resolution__prop,
> + revert_propset__dir,
> + revert_propset__file,
> + revert_propdel__dir,
> + revert_propdel__file,
> ]
>
> if __name__ == '__main__':
Re: [PATCH] fix for revert notification bug (1.4.0 regression)
Posted by Ed Price <ed...@gmail.com>.
> > Looks great to me, applied in r22051. Will propose for backport to
> > 1.4.x once I've confirmed that it works there. Thanks!
>
> Thanks! Unfortunately it looks like the testcases are failing
> on windows (only) according to buildbot. Looking at the logs:
[...]
I can't test on windows so I'm not sure this is it, but I suspect
it's actually because I didn't properly regexp-escape the expected
output, and windows paths have backslashes... Patch attached!
-Ed
[[[
Properly regexp-escape expected output in revert tests added in
r22051. Fixes spurious test failures on windows.
Patch by: Ed Price <ed...@gmail.com>
* subversion/tests/cmdline/revert_tests.py
(revert_propset__dir): Regexp-escape expected output.
(revert_propset__file): Same.
(revert_propdel__dir): Same.
(revert_propdel__file): Same.
]]]
Re: [PATCH] fix for revert notification bug (1.4.0 regression)
Posted by Ed Price <ed...@gmail.com>.
> > Please find the fix (including new testcases) attached.
[...]
> Looks great to me, applied in r22051. Will propose for backport to
> 1.4.x once I've confirmed that it works there. Thanks!
Thanks! Unfortunately it looks like the testcases are failing
on windows (only) according to buildbot. Looking at the logs:
EXPECTED STDOUT (regexp):
Reverted 'svn-test-work\working_copies\revert_tests-12\iota'
ACTUAL STDOUT:
Reverted 'svn-test-work\working_copies\revert_tests-12\iota'
EXCEPTION: SVNLineUnequal
Weird... They look equal -- and apparently *are* equal on
platforms other than windows. Some kind of newline issue??
I don't know but I'll look into it.
-Ed
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: [PATCH] fix for revert notification bug (1.4.0 regression)
Posted by Garrett Rooney <ro...@electricjellyfish.net>.
On 10/11/06, Ed Price <ed...@gmail.com> wrote:
> On 10/11/06, Ed Price <ed...@gmail.com> wrote:
> > 1. Please see thread here:
> >
> > http://svn.haxx.se/users/archive-2006-10/0540.shtml
> >
> > I expect my patch broke it, so I'll try to fix it! :)
> > Please confirm that I may file an issue.
>
> Please find the fix (including new testcases) attached.
>
> I suggest applying to trunk and nominating for 1.4.x backport,
> as it does fix a regression in 1.4.0 and seems quite safe.
>
> I'm happy to rev the patch if there are any issues with it.
> Please let me know!
Looks great to me, applied in r22051. Will propose for backport to
1.4.x once I've confirmed that it works there. Thanks!
-garrett
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org