Back in the days, SF was unreliable. I don't know the exact reason but i suspect that they
acquired more projects in a short time frame than their servers could cope with. The result
was that their cvs server was more down than up and mails took at times over a day to be
delivered over the mailinglists. The cvs issues were mitigated a bit by introducing a commit
only cvs server for developers and a second read only one for all the users. Although this
made development more reliable for the developers, the problems remained for the users and
they got the additional problem that commits would take a day to be visible to them.
As a result of these problems, first MPlayer and not much later FFmpeg moved to their own
server. At that time it was interesting to see, that after FFmpeg moved their cvs repo to the
MPlayer server, SF's cvs server became quite a lot faster. It might be coincidence, but still
funny to see.
I don't know whether SF still has these problems, i haven't used it for a very long time. But
i guess that they got enough hardware and a proper scalable framework to cope with it these