[Python-Dev] migration away from SourceForge?

[Tim Peters]

[Skip Montanaro]
>> Is it time to think seriously about moving away from SourceForge?

[Martin v. L=F6wis]
> Any proposal to move away from SourceForge should include a proposa=
l
> where to move *to*. I highly admire SourceForge operators for their
> quality of service, and challenge anybody to provide the same quali=
ty
> service. Be prepared to find yourself in a full-time job if you wan=
t
> to take over.

I'm not sure that better alternatives for *some* of what SF does coul=
dn't be
gotten with reasonable effort.  For example, on a quiet machine, I ju=
st did
a cvs up on a fully up-to-date Python, via SF.  That took 147 seconds=
.  I
also did a cvs up on a fully up-to-date Zope3, via Zope Corp's CVS se=
tup.
That took 9 seconds.  I expect at least as many (probably more) peopl=
e hit
Zope's CVS as hit Python's CVS, and ZC appears to put minimal effort =
into
maintaining its public CVS servers.  A crucial difference is that SF =
CVS has
to serve hundreds of thousands of people, and ZC's more like just hun=
dreds.

> SourceForge performance was *much* worse in the past, and we didn't
> consider moving away, and SF fixed it by buying new hardware. Give
> them some time.

There have been times over the past few weeks when cvsup time via SF =
was as
bad as it's ever been, meaning > half an hour to finish.  There have =
also
been times when it's been quite zippy.  I think they've made tremendo=
us
strides in cutting response time for the trackers, though (that was i=
ndeed
very much worse in the past).