Bug#513743: OOo calc misfunction

[Rene Engelhard <rene@debian.org>]

[ why were you dropping the submitter and the upstream people who might
  know waht the new gcc does wrong? ]

Hi,

Matthias Klose wrote:
> Apparently the build logs are not available anymore, and the package
> versions used to build the package are not exactly known.

Uhm, that's not true. I said them on IRC to you before you wrote this mail
(for ppc):

ppc:

-15: g++-4.3_4.3.2-1 gcc-4.3_4.3.2-1
-17: ++-4.3_4.3.2-4 gcc-4.3_4.3.2-4

i386 doesn't have logs available for -15 and .16 on buildd.d.o although built
by the buildd (yay for sucky buildds), but -17 is:
g++-4.3_4.3.2-4 gcc-4.3_4.3.2-4

>  - Rene confirms the segfault when rebuilt with 4.3.3-3 on amd64.

And I confirm that building against testings gcc works.

>  - Seems to be seen on more than one architecture.

At least i386, amd64, ppc. Didn't test other archs, but I guess it's safe
ro guess it happens everywhere when built with a new gcc.

>  - Downgrading the -core package to -15 avoids the crash.
>    - Does just using libuno_sal.so.3 from -15 avoids the crash
>       as well?

Didn't try.

>    - If yes, is it possible to rebuild libuno_sal.so.3 with
>      object files from the -15 build until the crash is avoided?

? I don't have the object files from the -15 build anymore. And
I think it's waste of time (and not really possible either) to build
-17 on testing, get the .os and put them into a -17 build on sid...
Especially not if using pbuilder.

>  - Is the crash seen when rebuilding with 4.3.2-4 / 4.3.3-1?
                                            ^^^^^^^

Those are not in Debian anymore, so not fetchable by pbuilder....

But for 4.3.2-4 see above, the broken builds by the buildds used 4.3.2-4.

Regards,

Rene