Seeking initial reactions: Moving setup from CVS to Subversion?

[Max Bowsher]

Igor Pechtchanski wrote:
> Well, since my name's been mentioned...  Actually, you did miss a whole
> category of people -- the anonymous CVS users...  But read on.

Well, there's always anonymous svn. Restricted to that role, the learning 
curve from cvs to svn is small indeed.

> Personally, I don't mind having to install another VCS to work on setup,
> especially since the usage transition seems relatively painless, and if
> only for the ability to do disconnected 'add's and 'diff's...  It would
> also give me an opportunity to play with subversion, learn it, and maybe
> consider it for my local version control on other projects.
>
> One thing, however, that did come up, and that I deem important, is that I
> sometimes use the web interface to CVS (the CVSWeb that CGF mentioned) to
> either track changes or retrieve an older version of some file.  That and
> the commit messages to cygwin-apps-cvs are pretty much essential for me,
> so unless they could be done with subversion as well (and as nicely), I'd
> have to speak out against the switch.

Both could.

> A wild idea: if there were some way to present a CVS view on a subversion
> repository (i.e., have an adapter client that behaves exactly like cvs),
> and to hook up that view to CVSWeb and anonymous read-only repositories
> (so that people wouldn't have to use svn unless they mean to make changes
> to setup), that would, IMO, satisfy pretty much everyone (including me).

It wouldn't be particularly hard to have svn commits replicating into a 
read-only cvs repository. I'll keep the idea in mind should the resource 
situation on sourceware ever change sufficiently that a pilot svn project 
becomes feasible.

Max.