Re: svn commit: r25204 - trunk/subversion/tests/cmdline

On Wed, 30 May 2007, Malcolm Rowe wrote:

> On Wed, May 30, 2007 at 08:20:23AM -0700, kameshj@tigris.org wrote:
> > +def copy_src_detection_bug_if_target_has_many_ancestors_in_same_commit(sbox):
>
> The 80-column line length limit is a limit, not a target :-)

Ugh, +1. Somewhat shorter, and definitely more succinct:

  detect_copy_src_for_target_with_multiple_ancestors

Or:

  detect_copy_src_when_multiple_potential_srcs_exist

...
> > + svntest.actions.run_and_verify_commit(wc_dir, expected_output,
> > + expected_status, None,
> > + None, None, None, None, wc_dir)
> > + saved_cwd = os.getcwd()
> > + os.chdir(A_copy_of_B_C_path)
>
> chdir() calls should be inside a try/finally block so that an exception
> inside won't break the rest of the tests (due to failure to reset the
> cwd).
>
> > + svntest.actions.run_and_verify_svn(None, [], [], 'merge', '-g')
> > + os.chdir(saved_cwd)
>
> You don't actually appear to be testing that the merge is actually doing
> anything? The comments above implied that the merge would fail to
> detect the right copy source - can you check that?

Yeah, expand this test to:

1) Test test the merge itself.

2) Test the svn:mergeinfo prop on the merge target. When we have
auditing, we can use that instead, but until then, just examine the
property and its value.