[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: [PATCH] Fix merging with broken softlink as target

From: Daniel Shahaf <d.s_at_daniel.shahaf.co.il>
Date: Mon, 12 May 2008 23:32:23 +0300 (Jerusalem Daylight Time)

Karl Fogel wrote on Mon, 12 May 2008 at 16:26 -0400:
> Karl Fogel <kfogel_at_red-bean.com> writes:
> > Hmmm. If the goal is to check for symlinks, not just specialness, maybe
> > it would be a good idea to check that the contents of the special file
> > begins with "link:", indicating a symlink? There may be other kinds of
> > special files in the future...
> >
> > Hmm, then again, it's highly likely that future kinds of special files
> > would want the same treatment anyway. So I've just adjusted the comment
> > a bit to talk about that; I'm running 'make check' now (paranoia, I know
> > you already did it) and will commit soon.
>
> Hmmm. I'm still getting a failure over ra_neon:
>
> prop_tests.py 26: test handling invalid svn:* property values
>

Karl, does it pass without David's patch? I only ran file:// checks
here...

(Normally I run either svnserveautocheck or davautocheck, chosen at
random)

Daniel

> When I say I run 'make check', what I really mean is I run the
> regression test suite over at least three RA layers:
>
> make check CLEANUP=true
> make svncheck CLEANUP=true
> make neoncheck CLEANUP=true
> make serfcheck CLEANUP=true /* Usually run this, not always */
>
> Now, the latter two, Neon and Serf, require that I start up the httpd
> server of course. I usually do that by hand, but 'make davautocheck'
> should automate the process if you want.
>
> The above failure is from 'make neoncheck'. Do you know what's causing
> it? (I haven't looked into it yet, having choosen instead to
> shamelessly toss the ball back into your court :-) .)
>
> Below is the exact patch I was testing, using approximately r31143 of
> trunk:
>
> [[[
> Fix "file not found" error when a merge target is a broken symbolic link.
>
> Patch by: David O'Shea <david.oshea_at_s3group.com>
>
> * subversion/libsvn_wc/merge.c
> (svn_wc__merge_internal): Diff against detranslated file if the
> target is "special".
>
> * subversion/tests/cmdlink/merge_tests.py
> (merge_broken_link): New test function.
> (test_list): Call the new test function.
> ]]]
>
> Index: subversion/libsvn_wc/merge.c
> ===================================================================
> --- subversion/libsvn_wc/merge.c (revision 31142)
> +++ subversion/libsvn_wc/merge.c (working copy)
> @@ -674,9 +674,18 @@
> }
> else
> {
> - svn_boolean_t same;
> + svn_boolean_t same, special;
> + /* If 'special', then use the detranslated form of the
> + target file. This is so we don't try to follow symlinks,
> + but the same treatment is probably also appropriate for
> + whatever special file types we may invent in the future. */
> + SVN_ERR(svn_wc__get_special(&special, merge_target,
> + adm_access, pool));
> SVN_ERR(svn_io_files_contents_same_p(&same, result_target,
> - merge_target, pool));
> + (special ?
> + tmp_target :
> + merge_target),
> + pool));
>
> *merge_outcome = same ? svn_wc_merge_unchanged : svn_wc_merge_merged;
> }
> Index: subversion/tests/cmdline/merge_tests.py
> ===================================================================
> --- subversion/tests/cmdline/merge_tests.py (revision 31142)
> +++ subversion/tests/cmdline/merge_tests.py (working copy)
> @@ -11258,6 +11258,7 @@
> expected_status.tweak('H_COPY/psi_moved', status='MM')
> svntest.actions.run_and_verify_status(wc_dir, expected_status)
>
> +
> #----------------------------------------------------------------------
> # Issue #3157
> def dont_explicitly_record_implicit_mergeinfo(sbox):
> @@ -11349,6 +11350,32 @@
> None, None, None, None, None, 1)
>
>
> +# Test for issue where merging a change to a broken link fails
> +def merge_broken_link(sbox):
> + "merge with broken symlinks in target"
> +
> + # Create our good 'ole greek tree.
> + sbox.build()
> + wc_dir = sbox.wc_dir
> + src_path = os.path.join(wc_dir, 'A', 'B', 'E')
> + copy_path = os.path.join(wc_dir, 'A', 'B', 'E_COPY')
> + link_path = os.path.join(src_path, 'beta_link')
> +
> + os.symlink('beta_broken', link_path)
> + svntest.main.run_svn(None, 'add', link_path)
> + svntest.main.run_svn(None, 'commit', '-m', 'Create a broken link', link_path)
> + svntest.main.run_svn(None, 'copy', src_path, copy_path)
> + svntest.main.run_svn(None, 'commit', '-m', 'Copy the tree with the broken link',
> + copy_path)
> + os.unlink(link_path)
> + os.symlink('beta', link_path)
> + svntest.main.run_svn(None, 'commit', '-m', 'Fix a broken link', link_path)
> + svntest.actions.run_and_verify_svn(
> + None,
> + expected_merge_output([[4]], 'U ' + copy_path + '/beta_link\n'),
> + [], 'merge', '-c4', src_path, copy_path)
> +
> +
> ########################################################################
> # Run the tests
>
> @@ -11515,6 +11542,7 @@
> server_has_mergeinfo)),
> SkipUnless(dont_explicitly_record_implicit_mergeinfo,
> server_has_mergeinfo),
> + merge_broken_link,
> ]
>
> if __name__ == '__main__':
>

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-05-12 22:32:38 CEST

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.