[MLton] Replacing Subversion with a DVCS

[Matthew Fluet]

As Ville inferred, I have been using "git svn" to stage some work. 
Probably for many of the same reasons that others have mentioned: moving 
work-in-progress between machines, checkpointing intermediate states 
(especially when there may be long stretches between opportunities to 
work), rebasing against trunk changes, ability to revise history (merge 
commits, etc.) before publishing on SVN, etc.

I think that Subversion is working well enough for the current level of 
development.  And I don't think changing the infrastructure is going to 
change the level of development significantly.  As Dan mentioned, there 
are some other research projects that are using MLton (Purdue!), but 
choose not to host their development in the SVN repository --- but that is 
a social problem, not a technical one.  I don't think that there has ever 
been a request for committer status that we have not granted.

So, while I think it makes sense to stick with Subversion for the time 
being, I don't think it prevents anyone from experimenting with DVCS in 
parallel.  That is, people should feel free to advertise git urls (either 
on the mailing list or on the wiki) for interesting branches that they 
wish to publish and maintain.  I don't think that is at odds with having 
the SVN/trunk be the "ground truth" and, as we've seen with "git svn", any 
changes that are merged into the trunk come along with their local 
history, etc.  (The only downsides that I've noticed are the rapid 
timestamps, the svn:ignore property, and some file renames/moves that 
don't get recorded as such in the SVN.)

I also have to admit that I don't quite "get" all the advantages of a 
DVCS.  Yes, having the entire history available locally (with no network 
traffic) is nice.  Also, merging trunk changes into a branch is pretty 
easy with git (but, recent versions of Subversion have much better merge 
support).  But, beyond that, I don't really get the workflow.

-Matthew

On Wed, 15 Apr 2009, Daniel Spoonhower wrote:
> Though I have yet to commit anything to the trunk, I also use "git svn"
> to access the repository and work with a branch.  I also think it's
> worth considering a DVCS: though the trunk is relatively stable, there
> are a number of research projects based on MLton and it would be a
> useful way of sharing/distributing that work.
>
> --djs
>
> Ville Laurikari wrote:
>> On Tue, Apr 14, 2009 at 05:23:09PM +0200, Wesley W. Terpstra wrote:
>>> On Tue, Apr 14, 2009 at 9:49 AM, Ville Laurikari <ville at laurikari.net> wrote:
>>>> It seems to me that the MLton project would benefit from a distributed
>>>> version control system.
>>> I don't think it's so important. MLton development is pretty slow at
>>> the moment and at least I've had no problems with using subversion.
>>
>> I wouldn't have asked if I didn't have any problems with using
>> Subversion.  Accessing mlton.org from some machines at work is quite
>> cumbersome.  Either I had to spend time setting up ssh tunnels to
>> access mlton.org or work without version control.
>>
>> Luckily, I've recently discovered git-svn which solves this problem
>> for me.
>>
>>> Don't fix it if it ain't broke?
>>
>> Depends on what constitutes "broken" and how much an improvement
>> costs :)
>>
>> Subversion is bad enough that I would argue that switching to
>> something better might make sense even for a one-foot-in-the-grave
>> project such as MLton.
>>
>> Anyway, this was just a probe to see if there's any interest in this
>> sort of thing.  There appears to be none, so never mind...
>>
>> --
>> http://laurikari.net/ville/
>>
>> _______________________________________________
>> MLton mailing list
>> MLton at mlton.org
>> http://mlton.org/mailman/listinfo/mlton
>>
>
> _______________________________________________
> MLton mailing list
> MLton at mlton.org
> http://mlton.org/mailman/listinfo/mlton
>