[SF]VM building project third report

Ian Piumarta ian.piumarta at inria.fr
Sun Jun 9 21:28:16 UTC 2002


On Sun, 9 Jun 2002, Ned Konz wrote:

> Is Ian's work on the Unix VM going to be merged with the SF stuff any 
> time soon?

Everything in my sources that is outside of platforms/unix is pulled
verbatim from SF.  Everything related to configure and make is contained
entirely within platforms/unix and is "immune" to anything outside that
tree.  (In particular, the "makeMakefileFragsAndProveFermatsLastTheorem"
scripts are irrelevant to [and ignore entirely by] my build process --
which I hope is now infinitely simpler than anything that grew out of
the original "unholy 3.1 mess".)

My goal has always been that my build process support cvs checkout of the
SF platforms/Cross tree over the top of (or as a bulk replacement for) the
one that comes with my sources (or vice-versa: replacing the SF
platforms/unix with mine), followed by a VMMake of platforms/src to arrive
at a fully-working set of Unix sources.

W.r.t merging: there was much discussion about a month ago concerning the
whole SF process, which led utlimately to violent agreement by all
concerned.  Goran then volunteered (with unanimous support) to write the
"SF Squeak Constitution" but the SF developers list has gone utterly
silent since then.  I would have no objection whatsoever to maintaining
the Unix port at SF according to the procedural concensus that was reached
during those discussions.

But for now, in the absence of any definitive "green light" from the SF
hackers, I have no intention of modifying/replacing any of the Unix code
on SF -- since I don't consider it as "my" code.  AFAIC it "belongs" to
the group of people who created it, and I am as reluctant to fiddle with
their code without permission as I would hope others would be reluctant to
fiddle with my code without permission.

In the interim I've been rewriting my Squeak web site and learning how
cvsweb works (mainly out of curiosity).  My Squeak sources are not
currently managed by CVS but as of this moment there is no technical
reason why I couldn't put them into my local CVS repository, make them all
publically browsable via cvsweb, and hand out CVS accounts to people who
have contributed significant chunks of code to them (so as to permit
shared maintenance).  But this would obviously be a highly undersirable
situation to arrive at.

> I'm looking forward to sound support...

I hope you have a pleasant surprise!  (As noted in README.Sound, the
rewrite of OSS was done using the ALSA compatibility modules which work
way better than any Linux native OSS driver that I've come across.  The
upside is that the sound code is far more efficient and leaves way more
CPU free for Squeak itself.  The downside is that some Linux native
drivers may not implement sufficiently [or even correctly] enough of the
OSS API for them to work well with Squeak.  I figure this is acceptable
because the future of Linux sound lies undeniably with ALSA and anyone who
still hasn't switched to ALSA could maybe consider Squeak as an excellent
excuse to do so.)

Just curious: how many people would be interested in (and would make use
of) native ALSA sound support (i.e., not relying on ALSA OSS compatibility
modules) in Squeak?

Ian




More information about the Squeak-dev mailing list