[SciPy-dev] SVN "corruption" / bzr-svn oddity?

[Pauli Virtanen]

Thu, 16 Apr 2009 16:41:33 +0900, David Cournapeau kirjoitti:
> Pauli Virtanen wrote:
>> The Git repository indicates that something like this occurred:
>>
>>        5655 5656 5657 5658 5659 5660
>> 	o----o----o----o----o----o----o 5661
>>          \___________________________/
>>
>> So, the revision 5661 is based on 5655.
> 
> You can see this from svn as well, using the -v option of log:
> 
> svn log -r -v HEAD http://svn.scipy.org/svn/numpy/trunk

Yes, if you know to look for it. But in Git it was obvious what happened, 
since its data model is simpler, so I looked first at it :)

[clip]
> The 'R' meaning replace. I don't know why bzr did this - it looks like a
> bzr-svn bug to me.
> 
> I don't know what's best. I tried your suggestion locally:
> 
> svn cp file:///usr/media/src/dsp/scipy_trans/yo/trunk@5660
> file:///usr/media/src/dsp/scipy_trans/yo/trunk
> 
> but this creates a new trunk directory inside trunk. Omitting the trunk
> for the copy destination does not work, svn complains about overwriting
> an existing directory.

Does

	svn rm file:///.../yo/trunk
	svn cp file:///.../yo/trunk@5660 file:///.../yo/trunk

work then?

> I don't know what's best: rewriting the svn repository (keep everything
> up to 5660 and then commit 5661 as Josef intended, i.e. just changing
> distributions.py), or doing the same by reverting the changes.

I'd rather try to restore the trunk tree to 5660 and commit on top of 
that; otherwise you can't do

	svn diff -r 5660:5661 test_basic.py

since SVN thinks 5660 lives in a different directory than 5661 although 
the directories have the same name...

Who said SVN is simple :)

-- 
Pauli Virtanen