I continued with the following experiments:
mkdir Test
svn add Test
svn commit
# repos revision = 20
svn delete Test
# no commit here, Test still exists in "deleted state"
# now tried to restore the repository to an earlier state
# ( just as an example, i know i could use revert in this special case )
svn cp -r 20 Test . # '.'
svn: 'Test' is in the way
# i didn't want to do an extra commit of the deleted 'Test'
# so i gave it a brute force delete...
rm -r Test
# and retried to get the revision
svn cp -r 20 Test . # '.'
# X # instruction at 0x77c42545 referenced mem at loc 0x0, read.
( using the 1.0.0 win32 binaries from subversion website on WindowsXP )
I guess this tells me two things:
- to retreive an older revision I have to delete it from the
server-repository-HEAD first,
and perform the copy at server side.
- accidentally losing a folder and/or .svn folder will give me a segfault.
----- Original Message -----
From: "Ben Collins-Sussman" <sussman@collab.net>
To: "Brian Denny" <brian@briandenny.net>
Cc: "C.A.T.Magic" <c.a.t.magic@gmx.at>; <users@subversion.tigris.org>
Sent: Saturday, February 28, 2004 19:14
Subject: Re: Why does "svn copy -r 28 http://rep/p/Test http://rep/p/Test"
produce a copy as Test/Test ? - but not if Test does not exist or is deleted
from HEAD before?
> 'svn cp A B/C' tests to see if C exists in B already. If it doesn't,
> then it creates C. If it does exist already, it creates B/C/A.
>
> And 'svn mv' follows the same rules, because it's the same as "svn cp;
> svn rm".
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Sat Feb 28 20:06:42 2004