[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#219928: debian-installer: build failures on sparc



stappers@stappers.nl (Geert Stappers) writes:

> On Tue, Nov 11, 2003 at 08:17:45PM +0100, Falk Hueffner wrote:
> > Joey Hess <joeyh@debian.org> writes:
> > 
> > > Goswin von Brederlow wrote:
> > > > > Then how do you know it is a bug in mklibs and it is not actually a
> > > > > missing symbol?
> > > > 
> > > > Which was the real reason it failed in every testcase I got my hand on
> > > > so far. All of those suposedly mklibs weak symbols bugs disapeared
> > > > when recompiling the faulty binaries against the proper versions of
> > > > libraries so far.
> > > 
> > > If I read between the lines right, this time it was breakage in brltty
> > > on sparc, and I've removed it from the pkg-lists for sparc now.
> > 
> > Does that mean we can close the bug?
> 
> Please see attached patch.
> 
> > 
> > -- 
> > 	Falk
> 
> Geert Stappers
> 
> diff -ur mklibs-0.1.12/debian/changelog mklibs-patched/debian/changelog
> --- mklibs-0.1.12/debian/changelog	2003-08-28 12:18:10.000000000 +0000
> +++ mklibs-patched/debian/changelog	2003-11-13 09:55:30.000000000 +0000
> @@ -1,3 +1,11 @@
> +mklibs (0.1.12.1) UNRELEASED; urgency=low
> +
> +  * Geert Stappers
> +    - Give tidy error on not found non-weak symbol;
> +    not a backtrace dump (closes: #211092, #219928)
> +
> + -- Geert Stappers <stappers@stappers.nl>  Thu, 13 Nov 2003 09:36:48 +0000
> +
>  mklibs (0.1.12) unstable; urgency=low
>  
>    * Alastair McKinstry
> diff -ur mklibs-0.1.12/mklibs.py mklibs-patched/mklibs.py
> --- mklibs-0.1.12/mklibs.py	2003-08-28 12:19:28.000000000 +0000
> +++ mklibs-patched/mklibs.py	2003-11-13 09:57:33.000000000 +0000
> @@ -466,7 +466,8 @@
>      for (symbol, is_weak) in needed_symbols.elems():
>          if not symbol_provider.has_key(symbol):
>              if not is_weak:
> -                raise "No library provides non-weak " + symbol
> +                print >> sys.stderr,"No library provides non-weak " + symbol
> +                sys.exit(1)
>          else:
>              lib = symbol_provider[symbol]
>              library_symbols_used[lib].add(symbol)

That might give a nicer looking error but doesn't affect the problem.

If brltty is used on d-i we need to add an option to keep some weak
symbols as a workaround to the reverse linking used.

MfG
        Goswin



Reply to: