[Bioperl-l] Possible migration to git/github
Chris Fields
cjfields at illinois.edu
Tue Apr 20 19:00:49 UTC 2010
On Apr 20, 2010, at 8:24 AM, Peter wrote:
> On Tue, Apr 20, 2010 at 11:20 AM, D. Joe Anderson <bioperl at etrumeus.com> wrote:
>> On Sat, Apr 17, 2010 at 11:18:13AM -0500, Chris Fields wrote:
>>
>>> I wanted to get the BioPerl community's general input on a
>>> possible Subversion to git/github migration for the BioPerl
>>> repository.
>>
>> Moving to git seems like a great good thing with few downsides.
>> Distributed revision control has been gaining ground for years,
>> and seems to have exploded with the development and adoption of
>> git. All the cool kids are doing it.
>>
>> Moving to github, however, brings up all the usual concerns with
>> getting locked-in to a proprietary application, since the github
>> software is not open source. Granted, given the decentralized
>> nature of git development, the code itself would be fine.
>> However, contributor metadata and workflows could be at risk
>> from changes in github's business model down the line.
>>
>> I realize these concerns face an uphill battle, given the
>> popularity of github in particular and of proprietary
>> software-as-a-service more generally (Google, various social
>> networking sites) and, most perniciously, the intrusion of
>> proprietary software-as-a-service into partially- or mostly-FOSS
>> projects (like Ubuntu One), but I felt they needed airing.
>
> Using git does not in anyway tie us to github.com - we could
> in theory host the "official" repository on the OBF servers or
> anywhere else (such as gitorious as you mentioned), and
> from a technical perspective this is easy. The authorship
> metadata would also be preserved (I don't understand your
> concern here).
>
> You are right there is a potential concern if the project comes
> depend on any of github's additions like the network diagram
> or commenting features, or using the github "pull request"
> notification system as part of a workflow to review code for
> merging. But a workflow is a social convention that can be
> changed as needed.
>
> I have no concerns over using git hosted on github from a
> vendor lock in point of view.
>
>> We should note here that Biopython is already using git and
>> github in some fashion:
>>
>> http://www.biopython.org/wiki/GitUsage
>
> Don't forget that BioRuby has been using github for even longer,
> as I pointed out near the start of this thread:
>
> http://lists.open-bio.org/pipermail/bioperl-l/2010-February/032353.html
> http://github.com/bioruby/bioruby
>
> Also Chris mentioned (off list) that Biolib is using github too:
> http://github.com/pjotrp/biolib/
>
> This does seem to confirm your observation of the popularity
> of github ;)
>
> Peter
> (Biopython)
I do share some of D. Joe Anderson's concerns, mainly re: use of github's other tools. However, I don't particularly see this as an insurmountable issue, just one that needs to be clarified w/ regards to how many of the github tools we intend on supporting. I don't envision a complete migration to their wiki services or bug tools for the time being.
And glad that Gitorious was mentioned, as it was another option I wanted to raise as well. It does have several large and very active projects (Qt, FreeBSD, etc), but very few (any?) bioperl devs on it, beyond me.
chris
More information about the Bioperl-l
mailing list