[rfk-dev] Moving off SourceForge
Peter A. H. Peterson
pedro at tastytronic.net
Sun Nov 4 08:50:26 PST 2012
Quoting Eric S. Raymond:
> Peter A. H. Peterson <pedro at tastytronic.net>:
> > Quoting Eric S. Raymond:
> > > There has been some talk of moving the project off SourceForge.
> > > On reflection, I don't think I understand why. In what respects
> > > does SourceForge seem to be a bad fit?
> > My motivation for doing so is that since most of the activity happens
> > on rfk.org, and we archive source code there as a record of all
> > rfkiana, sourceforge tends to become a bit forgotten.
> I think this calls for a change in mental habits. The git repo is your
> master record of rfkiana now, both code and NKIs. It will become even more
> definitively so when I fold in the two ancient releases and the Inform port.
I think the git repo is a much more comprehensive repo now, but I
wouldn't call it the master record of rfkiana. There are 30-some
ports on the site, many of which may not ever merge with git. There
are also other kinds of rfkiana, like "other appearances",
screenshots, etc. The git repo is now a much more comprehensive repo
-- thanks to you. But it's not yet the master record.
> > It's also a dependency on the project that generally serves little
> > purpose other than providing CVS.
> You're forgetting mailing lists. And the bugtracker. Forges really do
> serve purposes that aren't met by a bare repo - I think you're dismissing
> SF too quickly.
I provide the mailing lists. The bug tracker is no doubt a useful
feature, but the project is also small enough that I'm not sure we
*need* what SF offers. I'm not saying it's valueless, but it comes at
> > We forget our logins and then must go through
> > recovery processes. We have to wait 5 seconds to download tarballs
> > while looking at ads.
> In the brave new git world, you may never log into SF again, nor download
> a tarball. git clone is your friend! It makes these impediments go away.
Until we have to push a bugfix back up (and have forgotten our
password), or grant access to a new user (or, if we are using the
bugtracker, track bugs, etc.) If the repo was hosted locally, git
clone would still be our friend, but he'd be our roommate, not living
in another town.
The fact is that I have just always found dealing with SF.net kind of
a hassle, and the project is not so huge or busy that it requires the
(admittedly handy) infrastructure they provide.
> > On the other hand, by setting up a community rcs at rfk.org, we keep
> > everything under one roof, in one place. It's just cleaner and
> > streamlines the processes of updating code and/or including new ports.
> The process of updating code and/or including new ports is now "git push",
> and it matters little where the repo is actually located. That being the
> case, I don't see any case for moving it. Doing so would violate the
> expectations of people who come to SF expecting to find code.
If we moved the repo, I think we would probably deactivate the SF site
or just point them back to rfk.org. I don't think that the user
expectation of finding rfk code at SF.net is especially widespread or
necessary to protect.
That said, I don't think there's a compelling reason to move it right now,
so I think for now we should table this discussion unless we have to
make a decision that affects it (like relying on the bugtracker
heavily). We have enough other things to discuss, and only so much
Peter A. H. Peterson
Graduate Student Researcher
Laboratory for Advanced Systems Research
University of California, Los Angeles
More information about the rfk-dev