[petsc-dev] configure issues

Matthew Knepley knepley at gmail.com
Wed Feb 9 05:33:58 CST 2011

On Wed, Feb 9, 2011 at 4:45 AM, Jed Brown <jed at 59a2.org> wrote:

> On Wed, Feb 9, 2011 at 11:33, Matthew Knepley <knepley at gmail.com> wrote:
>> A single system, with real scoping, easy debugging, more readable code,
>> etc.
> Would you like a pony? What does this even mean? There is logic for a PETSc
> program to attach a debugger to itself. The details of that logic depends on
> tests that are run by configure. PETSc does not presently require Python at
> runtime. If you were to refactor it, where would the logic go and how would
> it be called from C?

If you refer to the first part of your comment above, it suggested
generating the correct code, and then questioned the utility of doing so:

What does this even mean?

  single system: Here I mean using just Python, rather than Python coupled
with the awfulness that is the C preprocessor

  real scoping: Python has nice scopes, rather than the underscore hell of
preprocessor names

  easy debugging: I can easily see the code I generated and debug it. This
is much more complicated with preprocessor logic

  readable code: See above

I cannot think of an argument for using the preprocessor.

What most experimenters take for granted before they begin their experiments
is infinitely more interesting than any results to which their experiments
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20110209/69039f26/attachment.html>

More information about the petsc-dev mailing list