dway at riskmetrics.com
Tue Feb 4 16:15:35 CET 2003
Ian Piumarta wrote:
> > > and possibly at least one pre-built Unix VM.
> > > (Which one(s)? Probably Linux at least.)
> I can do Linux (ppc/x86) and FreeBSD (x86) with no more hassle than typing
> `make' in a few different places.
That would be great.
> > > - Should we wait a few days between steps 1 and 2? This would let
> > > people test the final image a bit just to make sure there wasn't some
> > > major problem, and maybe check out any final cosmetic tweaks. Also, it
> > > would give the VM maintainers a few days to get their VMs ready. But
> > > perhaps this isn't necessary... maybe we should make sure the VMs are
> > > ready before we declare 3.4 final. (Certainly they should be almost
> > > ready anyway.)
> > Hmmm. The best would probably be to give the port maintainers a chance
> > to produce and verify new VMs. Otherwise we may end up with a port
> > needing some tweak to the image to work but the image was just marked
> > final... Ooops.
> > What do you say, port guys? (Tim, Ian, John, Andreas)
> Judging by the (lack of) recent commit activity I suspect the Mac/Win VMs
> are pretty stable. For Unix though it sounds like a safer plan to me,
> especially if a few people volunteer to test the VMs thoroughly. (In
> addition to several minor fixes that I've collected over the last couple
> of months there's a bunch of new FFI support code. OTOH, there's
> absolutely no obligation at all to pull this new code into the VM right
> now. We could be content with the few minor fixes and leave the rest for
> the ongoing major overhaul of the Unix code, which is due out in a week or
> two and is _huge_. Given the intention to announce things in public
> forums, a solid VM seems preferable to a much more capable -- but far less
> tested -- one.)
I tend to agree with the latter half of this, that it is probably better to go with the solid VM with a few minor fixes, and save the major changes for after the 3.4 release. People will still be able to run the newer VM with the 3.4 image later if they want.
> Speaking of which (which we were, even if it wasn't evident) is there
> anything in the 3.4 image w.r.t. VMM, Interpreter, et al., which is
> different to what's in a fully-updated TeaSqueak/Croquet image? I'm
> currently using the latter since it has BitBlt additions that aren't in
> 3.4 (or at least they weren't the last time I looked).
I don't have a strong opinion on this, although I suppose that Tim has a point that it's good for the released 3.4 VM to actually match the interpreter etc code in the image. But perhaps Unix VM builders haven't typically been building from the code in the image anyway. In any case, let's just decide on something. ;-)
> > > - Which Unix VMs should be available immediately upon the release of
> > > 3.4? Sources? A Linux VM? Debian and/or RPMs? Obviously we don't
> > Personally I think a source release from Ian is enough, as long as Ian
> > has verified that the most popular *nix VM works. That would be Linux.
> I have debian installed on both ppc and x86 machines, so I could easily
> build .rpms and .debs for these two architectures. I've never tried
> building a FreeBSD package so I don't know if it would be feasible or not
> to put one together in short order. (As an irrelevant data point, I'm
> developing/maintaining the Unix support code almost exclusively under
> Mac OS X these days.)
To summarize somewhat, I think at a minimum we could just require that the source release plus the Linux-x86 prebuilt VM are on the ftp site, in order to declare 3.4 available for Unix. If Ian can also build for ppc, .rpms & .debs, FreeBSD etc without too much trouble, that would also be great, but not absolutely required. Whichever are easiest. :-)
More information about the Squeakfoundation