[Openmcl-devel] How about Git?

Robert P. Goldman rpgoldman at sift.info
Mon Nov 30 08:14:11 PST 2015


On 11/30/15 Nov 30 -9:53 AM, Dmitry Igrishin wrote:
> 
> 
> 2015-11-30 16:40 GMT+03:00 Robert P. Goldman <rpgoldman at sift.net
> <mailto:rpgoldman at sift.net>>:
> 
>     With all due respect, this sounds nutty. If you have a patch, attach
>     it to the Trac ticket. You can make a patch using diff if it's too
>     hard for you to figure out svn diff. Or use git-svn so you can use
>     Git while the maintainers use svn.... 
> 
> I would consider this as a "workaround" for myself, because, whether I
> like it or not,
> the Git is de-facto standard VCS today. And I just don't see any wrong
> to migrate to
> the de-facto standard VCS, I can just say "it would be nice" :) But you
> may vote for
> status quo, of course.
> Anyway, its matter of taste which VCS to use. But I don't know anyone who
> wants to return back to SVN or CVS after the Git :)

Having used both, I would say that you are correct that git is the
standard *distributed* VCS today.

Having used both git and SVN, I appreciate the utility of git as a DVCS.
 I am *not* convinced that it is better than SVN as a CENTRALIZED VCS.
git submodules in particular are a hot mess (they couldn't be as solid
as SVN externals because git is a DVCS).

Also, when you first move from SVN -> git, there is a *huge* learning
curve (or, to put it differently, "denial of service").

So I think the question that the CCL maintainers should ask themselves
is: "would the additional help from outside developers compensate for
the cost of the SVN to git translation?"

Unless there is the prospect of a substantial increase in contributions,
you might find the translation to be a net loss.  That would be a
question you (the CCL maintainers) should ask yourselves based on how
many contributions you get now, and how many you can expect.

If there is only the occasional patch in prospect, it might be better to
steer outside developers who are so inclined to git-svn.

Gary's strategy of "if you care enough, you do the transition, and we
can decide whether to follow you or stay with svn" seems reasonable.  If
there isn't anyone who cares enough to do the transition, maybe there
aren't so many contributions in the prospect, either.

Cheers,
r





More information about the Openmcl-devel mailing list