You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Max Bowsher <ma...@ukf.net> on 2004/02/26 16:00:31 UTC
Scary dumper-loader bug
----------------- repro.sh -------------------------
#!/bin/bash
#./cvs2svn.py --dump-only test-data/split-branch-cvsrepos
rm -rf sb1.svn sb2.svn
svnadmin create sb1.svn
svnadmin create sb2.svn
svnadmin load sb1.svn < pathological-dump
svnadmin dump sb1.svn > intermediate-dump
svnadmin load sb2.svn < intermediate-dump
svnlook tree sb1.svn > sb1.tree
svnlook tree sb2.svn > sb2.tree
diff -u sb1.tree sb2.tree
----------------- END -------------------------
----------------- output --------------------------
--- sb1.tree 2004-02-26 15:43:38.611712000 +0000
+++ sb2.tree 2004-02-26 15:43:39.833468800 +0000
@@ -6,6 +6,7 @@
branches/
demo-node-0/
module/
+ branched-from-branch
branched-from-trunk
first_working/
module/
----------------- END -------------------------
In other words, THE CONTENTS OF THE REPO WAS CHANGED BY DUMP-LOAD.
The sometimes overly complex dumpfiles produced by cvs2svn seem to confuse
the loader into producing a repository which behaves correctly to "svn
checkout", but incorrectly to "svn log -v" and "svnadmin dump".
I'm attaching a bz2-ed copy of pathological-dump, so that the bug can be
reproduced without cvs2svn.
Max.
Re: Scary dumper-loader bug
Posted by Max Bowsher <ma...@ukf.net>.
Branko Čibej wrote:
> Max Bowsher wrote:
>
>> I'm attaching a bz2-ed copy of pathological-dump, so that the bug can be
>> reproduced without cvs2svn.
>>
> Wouldn't it be fairer to say that the bug is _in_ cvs2svn, and that it
> should be fixed to produce valid dump files? (Yes, there's obviously a
> bug in our loader in that it doesn't check the dumpfile contents
> carefully enough to produce an error, but that's another issue.)
I'd say there are two bugs:
* One in cvs2svn, which currently produces redundantly complex (but not
incorrect) dump files.
* One in the dumper-loader, which responds to redundant complexity by
producing garbage.
The dumper-loader one is what this email thread is about, because the
cvs2svn one is already in IssueZilla.
The file I provided is a reproduction for the dumper-loader bug.
Max.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Scary dumper-loader bug
Posted by Branko Čibej <br...@xbc.nu>.
Max Bowsher wrote:
>I'm attaching a bz2-ed copy of pathological-dump, so that the bug can be
>reproduced without cvs2svn.
>
>
Wouldn't it be fairer to say that the bug is _in_ cvs2svn, and that it
should be fixed to produce valid dump files? (Yes, there's obviously a
bug in our loader in that it doesn't check the dumpfile contents
carefully enough to produce an error, but that's another issue.)
--
Brane Čibej <br...@xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org