Re: Some improvements to cvs2svn (patches)

On 03/28/2011 09:29 AM, Michael Haggerty wrote:
> On 03/26/2011 12:46 PM, Robin Stocker wrote:
>> Am 24.03.2011 06:13, schrieb Michael Haggerty:
>>> I suggest that we write a few predicates for obvious things, like
>>> cvs_file_is_binary or even cvs_file_has_mode('b').  This would make it
>>> convenient for users to use without having to know what a lambda is or
>>> write a Python function.  Then it would make much sense to wrap the
>>> "KeywordHandlingPropertySetter('collapsed')" lines in the DVCS option
>>> files to make them only apply to non-binary files, which would make them
>>> a safer default than the present version.
>>>
>>> If you are interested, go for it.
>>
>> Done, see commit c42f007 in my repo:
>>
>> https://github.com/robinst/cvs2svn/commits/master
>
> Cool, committed as trunk r5331.

Thanks.

>>>> • Add option for excluding paths from conversion
>>> [...]
>> Ok, all of this is now in a5eab2d.
>
> Very nice.  Committed as trunk r5332.
>
> I edited your commits lightly for formatting; hopefully I haven't broken
> anything :-)

Works fine here :).

> I like the new property setters, because it kindof points the way to
> building up the property setters out of simpler building blocks.  I
> think several of the existing property setters could benefit from the
> same treatment.
>
> Are there any other cvs2svn rough edges that you had planned to work on?

Not at the moment, no. I'm now fighting with converting a hairy svn repo 
to git ;).

>   I've long had a bad conscience that the test suite does almost no
> testing of the DVCS backends; have you done any automated testing of
> your conversions to git?

No, just manual "diff -r" of checkouts.

-- 
   Robin Stocker

------------------------------------------------------
http://cvs2svn.tigris.org/ds/viewMessage.do?dsForumId=1667&dsMessageId=2715799

To unsubscribe from this discussion, e-mail: [[email protected]].