On Jan 25, 2008, at 08:06, Christopher Key wrote:
> I'm not sure if this belongs here or on the developers list, please
> let
> me know if it would be more appropriate elsewhere.
>
> Anticipating the upcoming release of SVN 1.5, I decided to have a play
> with the new relative externals, and almost immediately ran into a
> problem,
>
> # svnadmin create /tmp/repo1
> # svnadmin create /tmp/repo2
> # svn co file:///tmp/repo1 /tmp/wc
> Checked out revision 0.
> # svn ps svn:externals "external1 ^/../repo2" /tmp/wc
> svn: Invalid svn:externals property on '/tmp/wc': target '^/../
> repo2' is
> an absolute path or involves '..'
>
> After looking at the source code, it appears that the problem is the
> ambiguity introduced with relative externals, making it unclear
> whether
> you mean "URL DIR" or "DIR URL", and the code is assuming the former.
> Accordingly, if I instead do,
>
> #svn ps svn:externals "^/../repo2 external1" /tmp/wc
> property 'svn:externals' set on '/tmp/wc'
>
> then all is well.
That's how it's supposed to work.
> My point is whether this is the correct assumption to make. All the
> documentation I've seen uses externals in the form "DIR URL", e.g.
> http://svnbook.red-bean.com/en/1.0/ch07s03.html
That was the old externals format.
> and indeed I wasn't aware that the form "URL DIR" was even valid.
It wasn't until 1.5.
> Would it be sensible to
> standardise on one representation, silently assume this representation
> where there's ambiguity, and issue a deprecation warning where the
> external is unambiguously contradictory. My personal preference would
> be for "DIR URL", both because it's easier to read, and because that's
> what the documentation uses.
Read the release notes:
http://subversion.tigris.org/svn_1.5_releasenotes.html#relative-
externals
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: users-help_at_subversion.tigris.org
Received on 2008-01-25 20:47:26 CET