Bug#549407: ivtools 1.2.6-1 FTBFS on sparc

[Agustin Martin <agmartin@debian.org>]

retitle #549407 ivtools 1.2.6-1 FTBFS on sparc and powerpc
thanks

On Mon, Nov 02, 2009 at 10:28:07PM +0100, Agustin Martin wrote:
> I plan to upload a new NMU with minimal changes.
> 
> One is just calling the offending target (where things fail) with -j1. Seems
> that the only way to really test this in a equivalent box is a new upload
> (Yes, I hate this).
> 
> I also plan to fix lintian error 
> 
>  * E weak-library-dev-dependency
                                                                                                                                                                           
>           * ivtools-dev on libiv1 (>= ${binary:Version}) 
>           * ivtools-dev on libiv-unidraw1 (>= ${binary:Version})
> 
> Since I uploaded last NMU and changes are minimal, I will probably do this
> with no delay and soon. I will however wait a couple of days for
> suggestions/objections. Comments are of course welcome.

Uploaded (diff attached), but no luck with #549407. Grrr!!!

That is, #549407 is still present in powerpc and sparc (retitling to note
that). 

In the meantime I have sucessfully built ivtools 1.2.6-1+nmu2 in smetana, a
Debian sparc machine with no changes at all in the package. Just built fine
out of the box (nmu1 also did). So #549407 is really confusing. Noting that
smetana is a dual UltraSparc IIIi v210 while lebrun, the sparc autobuilder,
is a Dual UltraSparc III 750Mhz.

powerpc problem is exacly the same and did not appear for 1.2.6-1, only
appeared after 1.2.6-1+nmu1. However, nothing related to that was changed in
the nmu. The only change was that 1.2.6-1 was built in voltaire (a Dual G4
PowerPC 500Mhz) while 1.2.6-1+nmu1 was built in praetorius (IBM p690 LPAR, 2
x POWER4 1.3GHz). Everything else regarding compilation was exactly the
same.

I have looked a bit more at the build log and the supposedly missing files,
and they are non-dangling symlinks shipped in upstream tarball, seem no
missing at all. 

So, I am extremely confused about this bug. All the above seems to point to
a race condition, but I cannot find out where it happens.

I can put more -j1 in package debian/rules and see what happens when
uploading again, but I really dislike that. I also have the sparc debs built
at smetana that could be uploaded. Did not try in a powerpc box.

Comments are welcome,

Cheers,

-- 
Agustin