[Bioperl-l] SVN and ...Re: Perltidy

Chris Fields cjfields at uiuc.edu
Fri Jun 15 10:56:23 EDT 2007


On Jun 15, 2007, at 5:07 AM, Sendu Bala wrote:

>>>> ...
>>> Can we do any sort of massive conversion at some logical timepoint.
>>> Probably after a branch release or something?  Because it basically
>>> means we're going to have differences on nearly every line which is
>>> going to make diff-ing difficult when debugging old/new versions.
>>> Maybe it is not a problem because we aren't introducing and new  
>>> bugs!
>
> Sorry, can you clarify the problem you envisage? And why would  
> making a branch release help?

Maybe the worry is that mass conversion in such a large codebase  
could lead to hard-to-locate bugs.  Shouldn't occur but who knows w/o  
trying?

>> I agree; if we intend on doing this it should be all at once,  
>> maybe  on a branch dedicated to ensure that code changes don't  
>> tank tests  (they shouldn't but one never knows).  We would then  
>> need a script up- and-running that tidies everything up prior to  
>> commits (though what  happens if perltidy tanks?...).
>> Sendu, up for it?
>
> If its going to be difficult and a hassle, for such an unnecessary  
> thing I'm not sure its worth it. There are more pressing things to  
> be done for Bioperl.
>
> If I can just run perltidy on the entire package and commit, I'd do  
> it. If that's not appropriate, I won't.

The choices aren't necessarily all or nothing.  What about voluntary,  
recommended use of a perltidy config file included with the  
distribution, with additional 'caveats'?  See my response to Sean.

>>>> About svn
> [snip]
>> Stepped into that one, didn't I!  I'll look into how much effort  
>> is  involved and try getting something going in the next month or  
>> two,  maybe sooner if time permits.  I'm lacking on SVN-foo as  
>> well but it  might be worth looking into.
>
> I'd put this in the unnecessary-but-nice category as well. If it  
> will be as easy as my ->new change, go ahead. If not, there are  
> more pressing matters (POD fixing, test script updating and  
> finishing...).

A few other open-bio projects have actively discussed a CVS->SVN  
migration (BioRuby and I think BioPython, though the latter could be  
wrong).  As I said, "it might be worth looking into" to weigh the  
pros/cons, get others opinions from others who have made the  
transition, etc.  We could, as Jason suggested, even set up a tester  
SVN w/o making it the default codebase (lock it off to a few testers,  
have CVS commits automatically/manually carry over to SVN, etc).

I agree with you that it's not feasible to switch over prior to a  
release and that there are more pressing issues, but it doesn't hurt  
having an open discussion about it.

chris



More information about the Bioperl-l mailing list