On Sunday, December 8, 2002, at 09:53 PM, Martin Pool wrote:
> On 8 Dec 2002, Garrett Rooney <rooneg@electricjellyfish.net> wrote:
>
>> i believe you could do an 'svn revert added' to revert your add, then
>> do an 'svn up' to get the new version.
>
> No, that doesn't work.
>
> % date>added
> % svn add added
> A added
> % svn up
> svn: Obstructed update
> svn: failed to add file 'added': object of the same name already
> exists
> % rm added
>
> The previous command induces the bug, now I try to recover:
no, the bug is that svn up succeeds. it should have failed even though
'added' was moved. you should be required to do an 'svn revert added'
here before you can update. svn has no way to know that you moved
'added', so it shouldn't write the new file in it's place. debatably,
it should see that 'added' is no longer there, and get rid of the entry
for 'added' in the .svn/entries file, but it seems more consistent for
it to just error out until you resolve the situation in a way that
allows svn to know you've resolved it.
> % svn up
> A added
> Updated to revision 13.
> % ls -l
> total 4
> -rw-r--r-- 1 mbp mbp 29 Dec 9 13:47 added
> % svn stat -v
> 13 13 mbp .
> A 13 13 mbp added
> % svn revert added
> Reverted added
> ls
> % ls
> added
> % svn stat -v
> 13 13 mbp .
> ? added
> % svn up
> At revision 13.
> % svn stat -v
> 13 13 mbp .
> ? added
>
> This should be uptodate, not '?'.
>
> There isn't any obvious way to get back to a working state, which to
> me increases the severity of the bug.
oh, i agree it's a problem, but at this point we're just arguing about
where the problem is ;-) the issue (in my opinion) is that moving the
file aside should not have been enough to let 'svn up' complete.
-garrett
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Dec 9 04:03:57 2002