[petsc-dev] ugliness due to missing lapack routines

[Matthew Knepley]

On Thu, Feb 7, 2013 at 12:09 AM, Jed Brown <jedbrown at mcs.anl.gov> wrote:

>
> On Wed, Feb 6, 2013 at 11:00 PM, Matthew Knepley <knepley at gmail.com>wrote:
>
>> You are right. All of petscblaslapack.h should be generated :)
>
>
> Perhaps, but not in petscconf.h. I think rampant duplication is the much
> bigger problem. If we only had one place to modify to add a new interface,
> we'd be just fine.
>

I am fine with generating that header.


> Note that any changes to petscconf.h imply a _full_ rebuild of PETSc.
> Every single file. In contrast, only about 30 files need to be recompiled
> if you change petscblaslapack.h. Since configure is so hopelessly slow, be
> sure to make the generator callable without running all of configure.
>

There is a lot of stuff to do in configure :)


> I think generating this is a waste of time. The real solution (that would
> have saved millions over the years) would have been for LAPACK to always
> ship with a lapack.h declaring the interface.
>

The Demmel-Dongarra School of software engineering :)

    Matt

-- 
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/20130207/2288e081/attachment.html>