valgrind creeping into makefiles

[Matthew Knepley]

I am cool with removing it. It, of course, should also be configured, so it
would
be nice to put this in petscmpiexec in a way that reads config info.

  Matt


On 8/22/09, Dima Karpeyev <karpeev at gmail.com> wrote:
>
> I have one comment: "valgrind" sounds like a very tool-specific option.
> Are there other memory checking tools?
> Maybe there should be an option like "-memcheck [valgrind]" with an
> optional tool name,
> similar to "-start_in_debugger [gdb]"?
>
> Dmitry.
>
> On Fri, Aug 21, 2009 at 4:44 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:
>
>>
>>   I noticed valgrind commands creeping into PETSc example makefiles. I
>> believe this is a terrible model and would like to remove them.
>>
>>   I've added support for calling valgrind correctly from linux or the Mac
>> to $PETSC_DIR/bin/petscmpiexec as an optional first argument -valgrind.
>> Thus you can call any PETSc program with valgrind using, for example,
>> petscmpiexec -valgrind -n 2 ./ex1 -ksp_monitor
>>
>>   Now for makefile tests one can run, for example: make runex7
>> MPEXEC="petscmpiexec -valgrind" and it will run the runex7 test under
>> valgrind.
>> Is this an acceptable replacement for having all the horrible redundant
>> use of valgrind directly in the makefiles and do I have everyones permission
>> to remove the use of valgrind from the makefiles? If not, what support do I
>> need to add to get rid of the use of valgrind from the makefiles?
>>
>>
>>   Barry
>>
>
>


-- 
What most experimenters take for granted before they begin their experiments
is infinitely more interesting than any results to which their experiments
lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20090824/444b00fe/attachment.html>