On Thu, 22 Dec 2005, OpenMacNews wrote:
...
> > This sounds related to issue #2306, which was actually reported only
> > on Windows in the 1.3 release stream. The issue has some comments
> > about other OSes, but those are related to the working copy operation
> > logging code which was introduced on the trunk after 1.3 branched off.
> >
> > Here's the code snippet which is being run. The last line shown is
> > triggering the failure:
> >
> > client.propertySet(thisTest.getWCPath()+"/A/mu",
> > PropertyData.NEEDS_LOCK, "*", false);
> >
> > addExpectedCommitItem(thisTest.getWCPath(),
> > thisTest.getUrl(), "A/mu",NodeKind.file,
> > CommitItemStateFlags.PropMods);
> > assertEquals("bad revision number on commit", 2,
> > client.commit(new String[] {thisTest.getWCPath()},
> > "message", true));
> > File f = new File(thisTest.getWCPath()+"/A/mu");
> > assertEquals("file should be read only now", false, f.canWrite());
>
> thx for the heads-up.
>
> issue #2306 sez 'fixed' ... apparently not the case (at least for me ...)
And the "Target Milestone" says 1.3.x, which is somewhat odd, given
that the regression test wasn't backported to the 1.3.x branch.
Looking, I don't see that the fix was backported, either...if this
turns out to be the case, I'll propose it for backport to 1.3.1
(wouldn't hold up the 1.3.0 release for it).
> if i'm reading correctly, there should be a "/A/mu" in my build tree somewhere ... alas, nada.
It's probably cleaned up by the JavaHL test suite. As we discussed on
IRC, please try this backport of the regression test I'd committed to
trunk (patch to 1.3.x attached).
svn-1.3.x$ cd subversion/tests/clients/cmdline/
dlr@anvil:cmdline$ ./switch_tests.py 17
PASS: switch_tests.py 17: refresh the WC file system read-only attribute
--
Daniel Rall
- text/plain attachment: patch
- application/pgp-signature attachment: stored
Received on Sun Dec 25 04:04:17 2005