Just check-in changes to "mesh.cc", as usual. In v0.8.3, maybe I should 
make 'configure' download "mesh.cc" from the web...

>
> You have an incomplete BLAS.
>
>  

OK, maybe it is a problem with the host that ran the last build...

--Leif

On 5/10/07, Leif Strand Roundup Issue Tracker
<issue_tracker@geodynamics.org> wrote:
>
> Leif Strand <leif@geodynamics.org> added the comment:
>
> I'm renaming this issue, because two more PETSc problems popped up today.  I'm
> sure they'll be more.
>
> I just noticed that there are actually two problems in the build output sent-in
> by Tabrez.  There is VecScatter{Begin,End}, which was documented previously.

Yes, I wish they had told me.

> But also, the function SectionRealGetLocalVector is missing.

I did this. I changed the name to Create since that is what it does. I
forgot that
PyLith was using it. How do I push the fix?

> Further, the PyLith BuildBot (which I'm still working on) failed today due to a
> a PETSc configure problem:

You have an incomplete BLAS.

  Matt

> UNABLE to CONFIGURE with GIVEN OPTIONS    (see configure.log for details):
> ---------------------------------------------------------------------------------------
> Incomplete BLAS install? Perhaps blas package is installed - but
> blas-dev/blas-devel is required.
> *********************************************************************************
>
> http://www.geodynamics.org/buildbot/PyLith/
>
> ----------
> nosy: +baagaard, knepley, stali
> title: change to VecScatterBegin/End -> petsc-dev chaos torments PyLith
>
> _____________________________________________________
> Roundup issue tracker <issue_tracker@geodynamics.org>
> <http://geodynamics.org/roundup/issues/issue108>
> _____________________________________________________
>

I'm renaming this issue, because two more PETSc problems popped up today.  I'm
sure they'll be more.

I just noticed that there are actually two problems in the build output sent-in
by Tabrez.  There is VecScatter{Begin,End}, which was documented previously. 
But also, the function SectionRealGetLocalVector is missing.

Further, the PyLith BuildBot (which I'm still working on) failed today due to a
a PETSc configure problem:

UNABLE to CONFIGURE with GIVEN OPTIONS    (see configure.log for details):
---------------------------------------------------------------------------------------
Incomplete BLAS install? Perhaps blas package is installed - but
blas-dev/blas-devel is required.
*********************************************************************************

http://www.geodynamics.org/buildbot/PyLith/

Tabrez,

On Tuesday, the PETSc developers changed the order of arguments for 
VecScatterBegin and VecScatterEnd. Unfortunately, this breaks our 
newly-released PyLith v0.8.2.

Fortunately, the fix is easy. You can edit your copy of the code, moving 
the last argument of each call so that it becomes the first argument. 
Or, you can download a new, updated "mesh.cc". See the following page 
for details:

http://geodynamics.org/roundup/issues/issue108

--Leif

Tabrez Ali wrote:
> Leif
>
> While compiling Pylith I seem to get some errors which are attached 
> with the email. Btw I am using last nights tarball of petsc-dev which 
> I have built using gcc/g95/mpich  (as Intel 7x/8x/9x compilers wont 
> work at least on my setup (though they work fine with petsc-2.3.1 and 
> older) however Matt Knepley has said that he will be looking at this 
> issue soon).
>
> If you get a chance then please let me know as to what's wrong.
>
> Thanks
> Tabrez
>
> mesh.cc: In function `PetscErrorCode updateDisplacement(_p_SectionReal*,
>   _p_Vec*)':
> mesh.cc:435: error: `SectionRealGetLocalVector' undeclared (first use 
> this
>   function)
> mesh.cc:435: error: (Each undeclared identifier is reported only once 
> for each
>   function it appears in.)
> mesh.cc:437: error: cannot convert `_p_Vec*' to `_p_VecScatter*' for 
> argument `
>   1' to `PetscErrorCode VecScatterBegin(_p_VecScatter*, _p_Vec*, _p_Vec*,
>   InsertMode, ScatterMode)'
> mesh.cc:438: error: cannot convert `_p_Vec*' to `_p_VecScatter*' for 
> argument `
>   1' to `PetscErrorCode VecScatterEnd(_p_VecScatter*, _p_Vec*, _p_Vec*,
>   InsertMode, ScatterMode)'
> make[3]: *** [libPyLithMeshLibmodule_a-mesh.o] Error 1
> make[3]: Leaving directory `/tmp/stali/pylith3d-0.8.2/pylith3d/module'
> make[2]: *** [all-recursive] Error 1
> make[2]: Leaving directory `/tmp/stali/pylith3d-0.8.2/pylith3d'
> make[1]: *** [all-recursive] Error 1
> make[1]: Leaving directory `/tmp/stali/pylith3d-0.8.2'
> make: *** [all] Error 2

Fixed in r6831.

Users who have downloaded PyLith v0.8.2 can access the corrected 
"mesh.cc" here:

http://geodynamics.org/svn/cig/short/3D/PyLith/branches/pylith-0.8/pylith3d/module/mesh.cc

--Leif

This change causes the PyLith v0.8 build to fail. It looks like we may 
need to release v0.8.3 sooner than expected.

--Leif

if /usr/bin/g++-3.3 -DHAVE_CONFIG_H -I. -I. -I../.. -DMPICH_SKIP_MPICXX -I/home/buildbot/projects/PyLith/deps/x86_linux_single_nosched_gcc-3.3_g77-3.4_python-2.3_mpich2/petsc-dev -I/home/buildbot/projects/PyLith/deps/x86_linux_single_nosched_gcc-3.3_g77-3.4_python-2.3_mpich2/petsc-dev/bmake/linux-gnu-cxx-debug -I/home/buildbot/projects/PyLith/deps/x86_linux_single_nosched_gcc-3.3_g77-3.4_python-2.3_mpich2/petsc-dev/include -I/home/buildbot/projects/PyLith/deps/x86_linux_single_nosched_gcc-3.3_g77-3.4_python-2.3_mpich2/petsc-dev/include -I/home/buildbot/projects/PyLith/deps/x86_linux_single_nosched_gcc-3.3_g77-3.4_python-2.3_mpich2/petsc-dev/bmake/docsonly -I/home/buildbot/projects/PyLith/deps/x86_linux_single_nosched_gcc-3.3_g77-3.4_python-2.3_mpich2/petsc-dev/src/dm/mesh/sieve -I/home/buildbot/opt/mpich2/include -I/home/buildbot/projects/PyLith/deps/x86_linux_single_nosched_gcc-3.3_g77-3.4_python-2.3_mpich2/petsc-dev/externalpackages/Chaco-2.2/linux-gnu-cxx-debug/ -I/home/buildbot/projects/PyLith/deps/x86_linux_single_nosched_gcc-3.3_g77-3.4_python-2.3_mpich2/petsc-dev/externalpackages/Boost/ -I/home/buildbot/opt/python2.3/include/python2.3 -I/home/buildbot/projects/PyLith/install/x86_linux_single_nosched_gcc-3.3_g77-3.4_python-2.3_mpich2/lib/python2.3/site-packages/pythia-0.8.1.3-py2.3.egg -DALE_USE_DEBUGGING -DALE_HAVE_CXX_ABI -DALE_USE_LOGGING -DALE_LOGGING_USE_EVENTS -DALE_LOGGING_VERBOSE -I/home/buildbot/projects/PyLith/install/x86_linux_single_nosched_gcc-3.3_g77-3.4_python-2.3_mpich2/include -I/home/buildbot/projects/PyLith/deps/x86_linux_single_nosched_gcc-3.3_g77-3.4_python-2.3_mpich2/include    -g -O2 -MT libPyLithMeshLibmodule_a-mesh.o -MD -MP -MF ".deps/libPyLithMeshLibmodule_a-mesh.Tpo" -c -o libPyLithMeshLibmodule_a-mesh.o `test -f 'mesh.cc' || echo './'`mesh.cc; \
	then mv -f ".deps/libPyLithMeshLibmodule_a-mesh.Tpo" ".deps/libPyLithMeshLibmodule_a-mesh.Po"; else rm -f ".deps/libPyLithMeshLibmodule_a-mesh.Tpo"; exit 1; fi
mesh.cc: In function `PetscErrorCode updateDisplacement(_p_SectionReal*, 
   _p_Vec*)':
mesh.cc:437: error: cannot convert `_p_Vec*' to `_p_VecScatter*' for argument `
   1' to `PetscErrorCode VecScatterBegin(_p_VecScatter*, _p_Vec*, _p_Vec*, 
   InsertMode, ScatterMode)'
mesh.cc:438: error: cannot convert `_p_Vec*' to `_p_VecScatter*' for argument `
   1' to `PetscErrorCode VecScatterEnd(_p_VecScatter*, _p_Vec*, _p_Vec*, 
   InsertMode, ScatterMode)'
make[4]: *** [libPyLithMeshLibmodule_a-mesh.o] Error 1

-------- Original Message --------
Subject: 	change to VecScatterBegin/End
Date: 	Tue, 8 May 2007 22:19:44 -0500 (CDT)
From: 	Satish Balay <balay@mcs.anl.gov>
Reply-To: 	petsc-dev@mcs.anl.gov
To: 	petsc-dev@mcs.anl.gov

The prototypes for VecScatterBegin() and VecScatterEnd() are now
changed. This change is made to have these routines consistant with
PETSc calling convention [i.e the first argument should be the
appropriate object for that functionality]. i.e:

PetscErrorCode VecScatterBegin(Vec,Vec,InsertMode,ScatterMode,VecScatter);
to
PetscErrorCode VecScatterBegin(VecScatter,Vec,Vec,InsertMode,ScatterMode);

Satish