[VETO] old ws -- packages, define-behavior

[Michael Sperber]

>>>>> "Stephen" == Stephen J Turnbull <stephen at xemacs.org> writes:

Stephen> Non sequitur.  I'm asking for an incremental improvement in
Stephen> communication, and you're saying "you can't have that because we won't
Stephen> get a full-featured VC system with flexible merging."]]

Sorry---that wasn't clear to me.  I've asked Eric to send something in
soon.  (He actually keeps a blog of his project.)  Marcus is almost
ready to go.

Stephen> Read what you just wrote, and tell me that doesn't sound like "Ben
Stephen> _seems_ to have gotten better, but I'm not going to trust it because
Stephen> you know he used to be like that."

I'm confused---there was an intermittent period where things *were*
getting a little better (in 2003), but now, that seems to be over.
I'm not just referring to the patches, but to Ben's statements.

ms> Look at the FreeBSD logs for a good example.

Stephen> Can you point me to a single task that involved 10,000 lines or more
Stephen> of changes?  I'd like to see how they handled that.  (Presumably it
Stephen> wasn't a single patch, I mean a set of patches with a common goal of
Stephen> implementing a feature or completely refactoring a subsystem.)

They did it with lots of things, notably with the fine-grained SMP
support from 4.x from 5.x.  That was a massive multi-year project.
They were extremely careful factoring the changes into lots of little
steps.

Stephen> BTW: where are they?  The "sys" tree I have checked out doesn't have
Stephen> any ChangeLog files.

They keep the change logs via the mailing list.

http://lists.freebsd.org/pipermail/cvs-all/

Stephen> Maybe it's time to get a better revision control system.  I'll talk to
Stephen> the leading candidates for hosts (Tux and SunSITE) and see if they
Stephen> have experience or bad vibes with the usual suspects (Bitkeeper,
Stephen> subversion, darcs, arch).  I'm inclined against subversion; I imagine
Stephen> that it _would_ work for XE

That would help, but it won't solve all of the problem.  (I can offer
pretty extensive Subversion expertise, BTW.)  I'd be happy with a
commit-first approach, if we get religious about actually backing out
changes people disagree with.

-- 
Cheers =8-} Mike
Friede, Völkerverständigung und überhaupt blabla