[Biopython-dev] Subversion Repository

[Chris Lasher]

On 3/10/07, Michiel de Hoon <mdehoon at c2b2.columbia.edu> wrote:
> Chris Lasher wrote:
> > I hope the BioPython developers will consider a move to Subversion
> > seriously. If there is support from the devs, but no interest on
> > anyone's part to make it happen, given the proper people to contact, I
> > will be happy to get this moving as a way of contributing back to the
> > BioPython community.
>
> I know very little about CVS and Subversion (which is why I didn't
> respond to your original post). But I did notice that a lot of software
> projects are using Subversion instead of CVS nowadays, including Python
> itself. So I don't have any objections against Biopython moving to
> Subversion as well (unfortunately, I cannot be very help here either).
>
> If we are moving to Subversion though, I'd like to ask you not to make
> any changes until the next Biopython release comes out, which will be in
> about one week from now.

Since no one else has volunteered, I'm taking up responsibility for
the transition. I got the ball moving by contacting "support at
open-bio.org" to get alert them of our interest and get any contacts
we'll need to make this happen. Also, if anybody on the list has any
information that would be helpful in this (e.g., who administers the
CVS repo) please feel free to send it along. Likewise, feel free to
raise any questions, concerns, and comments on the list.

Once the Subversion repository is in place and in sync with the CVS
repository (done through cvs2svn), we have two options with regards to
the CVS repository:

1) Drop support for CVS.

What this means:
The CVS repository will either be shut down or be left but not
supported with updates to the code.

Avantages:
* Transitioning users to Subversion should be trivial due to
Subversion's inheritance from CVS.
 - Transition documentation already exists from svnbook.red-bean.com,
specific examples can be written on Biopython wiki, etc.
* Clean, less challenging solution.

Disadvantages:
* We will need to publicize in a big way that we've transitioned.
* Automated scripts on remote machines that depend on the CVS
repository would break or work with Biopython code that becomes
increasingly out of date.
 - Trivial to remedy (e.g., s/cvs up -dP/svn up/g)
* Obstinate users will complain.
 - We can't please everybody.

2) Allowing legacy support for CVS via the method found at:
<http://sam.zoy.org/writings/programming/svn2cvs.html>

What this means:
Briefly, the commits to Subversion repository are mirrored in the CVS
repository. CVS access becomes read-only, commits are not permitted.

Advantages:
* Allows legacy users of CVS repository to receive updates.

Disadvantages:
* We may not have enough administrative access to do this.
* This will require much more time to implement, test, and triple-check.
* Has anybody on the list ever done this? It could lead to a lot of
"learning experiences".

Questions, concerns, and comments welcome.

Chris