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

request autobuild attempt for gerris

Dear buildd admins,

I have an odd situation with the autobuild of gerris, failure in
unstable but success in experimental, for almost the same code. 

gerris uses mpi for parallelisation, so it build-depends on
mpi-defaults-dev. mpi-defaults-dev is a generic mpi pseudopackage which
depends on openmpi where available.  Where openmpi is not available we
have previously had it use lam4, but we are currently switching over to
mpich2 instead.

gerris in unstable has always failed to build on armel, mips, mipsel,
s390 and sparc, where mpi-defaults-dev 0.6 uses lam rather than openmpi.
The last failure (gerris 20110329-dfsg.2-1) was logged in April, see
https://buildd.debian.org/status/logs.php?pkg=gerris.  The symptom of
the build failure was:

> /bin/bash ../libtool --silent  --tag=CC   --mode=compile mpicc -DHAVE_CONFIG_H -I. -I../../../src -I.. -I../../.. -I/usr/include -DG_LOG_DOMAIN=\"Gfs\" -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -I/usr/include -DGFS_MODULES_DIR=\"/usr/lib/gerris\"   -DFTT_2D=1 -g -O2 -Wall -Werror-implicit-function-declaration -Wmissing-prototypes -Wmissing-declarations -pipe -std=c99 -D_XOPEN_SOURCE=500 -c -o libgfs2D_la-utils.lo `test -f 'utils.c' || echo '../../../src/'`utils.c
> ../../../src/utils.c: In function 'compile':
> ../../../src/utils.c:557: error: expected ')' before 'MODULES_FLAGS'
> ../../../src/utils.c:558: warning: too few arguments for format
> mpicc: No such file or directory

which only occurred alongside lam, not on the openmpi arches.

I assumed the build failure would be fixed with the switch to mpich2
(with mpi-defaults-dev 1.0 currently in experimental).  I uploaded a
test version of gerris to experimental to check, but the results are
odd. The buildds on your arches are not yet using mpi-defaults-dev 1.0
with mpich2, they're still using mpi-defaults-dev 0.6 with lam4.
Nevertheless, the build proceeded successfully using lam.  The lam
version is the same, 7.1.2-2.  The code in gerris was not changed.

It's not clear to me why the build failure with lam in unstable is no
longer occurring in experimental.  So I thought it would be appropriate
to ask if you can trigger a new build attempt for gerris in unstable
(20110329-dfsg.2-1), for  armel, mips, mipsel, s390 and sparc.

Or if you have any other ideas for what the difference in build success
might be, then let us know!



Reply to: