Re: Proposal: Subversion Migration

[MJ Ray]

Gregory John Casamento <greg_casamento@yahoo.com> wrote:
> Ummm.  This probably just means that he forgot to hit reply-all, a common
> mistake which I'm sure we've all made.   No need to be snippy. ;)

At some point, he put discuss-gnustep back into the to-list without
warning. Pretty rude and nothing to do with reply-all.

[...]
> * Actively developed, CVS hasn't seen many releases lately

Did you check your claim?

CVS's latest releases were on 3 October 2005 (feature development)
and 28 September 2005 (stable).

By comparison, SVN's latest release was on 25 August 2005 and
they seem to only have one series releasing at the moment.

> * Versions directories, file meta data, etc.
> * Handles file renames without the need to remove and re-add a file to the
> repo.
> * Atomic commits.  In CVS, if an error occurs during commit, it's likely that
> some changes made it and others didn't.  A bad situation.
> * Efficient handling of binary files.

Yes, I agree these have benefits, but they are features common
to most of the modern version control systems. My point is that
SVN is not the best of the modern ones and doesn't offer that
much to gnustep (for example, how many files are moving around?
Are binary files that much of the repository and do they change
often enough to be a pain?).

The biggest benefits for gnustep are likely to be through
distributed development - because developers are scattered all
over the world - and integration with the development tools.

It is relatively easy to migrate out of CVS, but migrating out
of SVN seemed far less simple, last time I looked.

> * Parseable output makes scripting easier.

There are tools which parse CVS output, such as cvs2cl, and
the formats are fairly stable. There are fewer tools working
with SVN, as far as I can tell. That's a function of age not
ease, but there's not much to choose between most version
control systems.