Re: Thread+GC issues on ARM

[Rob Browning]

Neil Jerram <address@hidden> writes:

> First of all, thanks for pushing 2.0.5 into Debian so quickly!

After a decent bit of work last year, I integrated my Debian development
directly into the Guile git tree (including much of the older history).
That, the fact that I don't have to split 2.0, and git-dpm/git-dch have
made things like shifting to a new version *much* easier.

> Out of interest, what are your criteria for dropping 1.8?

I want guile-2.0 to build (and pass make check) on all of the
architectures that guile-1.8 does:

  https://buildd.debian.org/status/package.php?p=guile-1.8&suite=sid

And I want all of the (relevant) packages that depend on Guile to work
with guile-2.0.

I think we're pretty close now.  Most of the remaining arch problems
(aside from arm) may be GC/timing related test failures, and I'm
planning to just mark those UNRESOLVED.

> Is / should there be any consideration for continuing to provide
> guile-1.8 as a legacy programming environment?  Personally I'd feel
> comfortable if the answer to that was "no" - because I think it's
> positive overall to concentrate everyone's mindshare on the present
> and future, and because I'm not aware of any significant support
> having been lost - but just wondering.

My impression is that we wouldn't be losing much by dropping 2.0, but
feel free to correct me.  The thing that makes me more inclined to drop
1.8 is a Debian specific issue -- guile-1.8 has never (and will never)
have thread support, and that seems like a fairly significant
limitation.

> I don't know.  Would you like me to provide you with my current libgc
> and guile changes, which give a successful build on armel, so that you
> can see if they work on armhf too?

Given your next message, I think I might just wait for a fixed Debian
libgc.  We can't really proceed without that anyway.

Thanks
-- 
Rob Browning
rlb @defaultvalue.org and @debian.org
GPG as of 2002-11-03 14DD 432F AE39 534D B592 F9A0 25C8 D377 8C7E 73A4