A quick summary of using cvs2svn.py, revision 6483 against our full
repository & pruned examples:
1) The bug fixed in rev #6418, that should fix (privately sent)
attachment "wrong-tuple.tar.gz" does indeed seem to have fixed the problem.
Original problem reported as:
http://www.contactor.se/~dast/svn/archive-2003-07/0504.shtml
2) The bug fixed in rev #6482, that should fix (privately sent)
attachment "no-attribute-keys.tar.gz" does indeed seem to have fixed the
problem, as hoped for in issue #1417.
Original problem reported as:
http://www.contactor.se/~dast/svn/archive-2003-07/0886.shtml
3) The python-trace problem(s) indicated by (privately sent) attachment
"tag-finishing.tar.gz", has not been fixed.
Original problem reported as:
A private email to Karl
Message Sent: Jul 15 00:37:35 GMT+1
Message Id: <3F133EB7.7070504@trouble.org.uk>
Subject: Re: cvs2svn.py: Similar trace to issue #1409 without --trunk
But I've re-created the important parts here...
A) I still get the python trace during the finishing phase, looking
something like:
$ rm -rf dev; python cvs2svn.py -s dev --create cvsroot/j_smsgw
...
finishing 'MM_030114_1335' as tag
finishing 'MM_021220_1455' as tag
finishing 'MERGEDFROM_MM_021111_1209' as tag
finishing 'MM_030325_1730' as tag
Traceback (most recent call last):
File "cvs2svn.py", line 2231, in ?
main()
File "cvs2svn.py", line 2227, in main
convert(ctx, start_pass=start_pass)
File "cvs2svn.py", line 2106, in convert
_passes[i](ctx)
File "cvs2svn.py", line 2070, in pass4
sym_tracker.finish(dumper, ctx)
File "cvs2svn.py", line 1692, in finish
self.fill_tag(dumper, ctx, name, [1])
File "cvs2svn.py", line 1639, in fill_tag
self.fill_name(dumper, ctx, tag, 1, jit_new_rev)
File "cvs2svn.py", line 1626, in fill_name
is_tag, jit_new_rev)
File "cvs2svn.py", line 1537, in copy_descend
next_src, next_dst, is_tag, jit_new_rev)
...
File "cvs2svn.py", line 1537, in copy_descend
next_src, next_dst, is_tag, jit_new_rev)
File "cvs2svn.py", line 1512, in copy_descend
dumper.copy_path(src_path, parent_rev, copy_dst, val)
File "cvs2svn.py", line 962, in copy_path
entries)
File "cvs2svn.py", line 663, in change_path
for ent in new_val.keys():
AttributeError: 'NoneType' object has no attribute 'keys'
That attachment "tag-finishing.tar.gz" holds three files that were
required to show the problem (There wasn't one single file it would fail
on).
If you remove "cvsroot/<blah>/FileA" from the CVS repository, all goes
smoothly in the conversion.
If, instead, you remove "cvsroot/<blah>/FileB", you get the same behaviour.
Both fileA and fileB have the "branches rooted in a dead revision"
problem - but that dead revision is version 1.1. Comments in the CVS
repository file seem to indicate that the file was actually created on a
branch, before being merged back into the trunk.
B) If, instead of either of those, you remove "cvsroot/<blah>/FileC",
you get a different problem, that is obviously related to the two files
above:
$ rm -rf dev; python cvs2svn.py -s dev --create cvsroot/j_smsgw
...
----- pass 4 -----
committing: Mon Mar 3 14:55:35 2003, over 0 seconds
deleting 1.1 : trunk/<blah>/fileA
no new revision created, as nothing to do
committing: Mon Mar 3 14:55:35 2003, over 0 seconds
deleting 1.1 : trunk/<blah>/fileB
no new revision created, as nothing to do
committing: Mon Mar 3 14:55:36 2003, over 0 seconds
No origin records for branch 'branch_mm_1-00_maint'.
Which is exactly the branch these 2 files were created on.
A copy of the data is winging its way to Karl separately.
Cheers,
Mike
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Jul 17 06:11:39 2003