[petsc-dev] prefix install and module file

[Matthew Knepley]

On Wed, Jan 14, 2015 at 12:59 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:

>
> > On Jan 14, 2015, at 12:03 PM, Aron Ahmadia <aron at ahmadia.net> wrote:
> >
> > Ah, I understand what you're doing now.  You're trying to have `pip`
> install the modules file to the correct location.
> >
> > At your top-level setup.py file you're overriding install():
> >
> >
> https://bitbucket.org/petsc/petsc/src/8f5bdbf243df394d73fcb8729f6a9331011281dc/setup.py?at=next#cl-157
> >
> > The setup.py build command doesn't take a prefix argument, so you
> shouldn't build/dump anything that is hardcoded to the prefix there (since
> as you say, you don't have it).
> >
> > You must defer that module dump command until you're installing (it
> doesn't actually make much sense for an in-place/development build
> anyway).  Is that hard to do?
>
>    We can always special case dumping in the install but the general PETSc
> install process is based on prefix during configure so you might have to
> start special casing the delay of more and more things. Essentially pip
> would restrict us to a configure ; make; make install that only accepted
> prefix at the install time. Do we want to be stuck with this restriction?
> Pip is new and evolving why do they still cling to prefix only at install
> time?
>

Definitely not. This is a senseless restriction, and ignores by far the
greater number of packages that conform to the GNU standard.

   Matt


>   Barry
>
>
> >
> > A
>
>


-- 
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/20150114/4e9bf624/attachment.html>