GNU bug report logs - #36747
Official MesCC bootstrap binaries differ from my locally built ones

Previous Next

Package: guix;

Reported by: Mark H Weaver <mhw <at> netris.org>

Date: Sat, 20 Jul 2019 22:46:01 UTC

Severity: serious

Done: Mark H Weaver <mhw <at> netris.org>

Bug is archived. No further changes may be made.

Full log


View this message in rfc822 format

From: Mark H Weaver <mhw <at> netris.org>
To: Jan Nieuwenhuizen <janneke <at> gnu.org>
Cc: 36747 <at> debbugs.gnu.org
Subject: bug#36747: Official MesCC bootstrap binaries differ from my locally built ones
Date: Mon, 12 Aug 2019 00:11:08 -0400
I wrote earlier:

> There are two bootstrap tarballs that differ:
>
> (1) static-binaries
> (2) guile-static-stripped
>
> In this message, I'll address only the 'static-binaries'.
>
> The only difference in the static-binaries is bash.  It turns out that
> the bash-4.4 configure script produces an incorrect compile-time
> configuration when built on Linux 5.x or later.  It boils down to this
> code in Bash's configure.ac:
>
> linux*)         LOCAL_LDFLAGS=-rdynamic          # allow dynamic loading
>                 case "`uname -r`" in
>                 2.[[456789]]*|[[34]]*)  AC_DEFINE(PGRP_PIPE) ;;
>                 esac ;;

I pushed a fix for this issue to the 'wip-cu-binaries' branch, commit
eefabc1db04c91d6954306e319820cd95190c25d.  With this fix, my
'static-binaries' tarball now matches yours.

What remains is to make 'guile-static-stripped' deterministic.  For now,
I suspect it might be sufficient to build it with #:parallel-build #f,
although of course it would be good to eventually fix the parallel build
to be deterministic.

It would also be good to enable building the bootstrap binaries from a
commit that's somewhere along the linear history of the 'master' branch.

To be continued...

     Mark




This bug report was last modified 5 years and 317 days ago.

Previous Next


GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.