[Biopython-dev] Biopython SVN Transition

[Chris Lasher]

On 10/30/07, Peter <biopython-dev at maubp.freeserve.co.uk> wrote:
> I was wondering if anyone would start suggesting moving to git or
> something else ;)

I tried Git and didn't like it. Bazaar suits me much better, and it
even has support for SVN repositories with bzr-svn. Git is not truly
cross-platform. It performs terribly on Windows. This left me looking
at Mercurial (Hg) and Bazaar (bzr). I liked the direction that Bazaar
was moving in and their emphasis on testing with real unit/regression
tests. For those interested, you can see some of the "literature" I
read through on my del.icio.us page:
http://del.icio.us/gotgenes/dscm

> Another thought; will existing developer accounts "just work" on the SVN
> system?  Also do you (Chris) have CVS access, and if not do you need or
> want it?

The existing developer accounts will "just work" because they're going
to do SVN over SSH. I have SSH access on the machine and CVS access as
well. Thanks for checking.

> > Also, I need to know what is needed from me in terms of helping all
> > the devs migrate to SVN. I produced a screencast series on Subversion
> > at http://showmedo.com/videos/series?name=bfNi2X3Xg and there is a
> > transition guide at http://svnbook.red-bean.com/en/1.4/svn.forcvs.html
>
> Sadly that didn't play with gnash 0.8, and I don't have Adobe's Flash
> plugin working on my 64bit Ubuntu.  I'll have to check that out on
> Windows later in the week :)

Bummer! Does nspluginwrapper not work?

> If you are able to field any queries on the mailing list, that would
> probably be fine.

I'd be happy to do that.

Should this page be renamed to SVN to be in the same line as tho CVS page?

> > Would providing links to these on the wiki be sufficient?
>
> If you could look after that aspect of the wiki, that would be great.

At some point I had started this:
http://biopython.org/wiki/Subversion_migration

> > What further information would you like to know? Subversion is not a
> > radical departure from CVS and many of the commands are a one-to-one
> > mapping. The biggest difference is commits occur for the whole
> > repository, not on a per-file basis, and directories are tracked, as
> > well.
>
> The fact the CVS and SVN are relatively similar is probably one reason
> why no-one has raised any real objections to the move.
>
> > Let's get a discussion on this and set a date soon.
>
> In terms of timing, how long do you/the OBF guys expect the transfer to
> take?  And would they prefer to do this over a weekend or mid week?

Not sure, let me ask Jason Stajich.

> Barring any problems with Biopython 1.44 which would force us to do
> another release in the very short term, I guess in the next fortnight is
> reasonable (especially if we only expect a couple of days downtime).

I think we could expect less than a full day downtime.

> Of course, I personally want to start working on the Seq objects and
> alignments - and Tiago wants to get back to his Population Genetics module.

By all means, continue using CVS until I get a firm date for the
Biopython Devs. Even if you have uncommitted changes when the CVS
server goes down, you can simply copy the files to your checked out
copy of the SVN repository and continue as is.

> P.S. Would you or any of the people doing the transition be able to sort
> out bug 2363?
> http://bugzilla.open-bio.org/show_bug.cgi?id=2363

That's a very good question. I wonder if cvs2svn is capable of picking
up those errors in commits and choose the proper format. I had trouble
getting a hold of an expert who could tell me how to identify files
committed as binary files, and how to change that to text (or vice
versa). I should send an email to the Subversion mailing list,
perhaps, or the CVS list if it's still active. I'll also check to see
if Jason knows.