[petsc-dev] [petsc-maint] valgrind detected on /usr/local/include and enabled but header files from this dir are not added to build path

Matthew Knepley knepley at gmail.com
Tue Oct 29 17:40:26 CDT 2013


On Tue, Oct 29, 2013 at 5:33 PM, Satish Balay <balay at mcs.anl.gov> wrote:

> On Tue, 29 Oct 2013, Satish Balay wrote:
>
> > On Tue, 29 Oct 2013, Matthew Knepley wrote:
> >
> > > > Matt is the one with vision about BuildSystem dependencies so I'll
> leave
> > > > it to him to comment.
> > >
> > >
> > > I see fooling with dependencies as a good thing in general to simplify
> > > things, but having not
> > > much to do with what Jed wants. We can easily buffer till kingdom come.
> > >
> > > What needs to be done is to overload write() for the logger, so that
> if no
> > > file is open it buffers,
> > > and then when a file is set, the contents get dumped in.
> >
> > The issue is: if there is an error in configure thats relavent [for
> > which we want the user to send us configure.log from the predetermined
> > location] - but this error happens before the 'file is set' i.e before
> > PETSC_ARCH is determined.
>
> I make some changes in 'balay/simplify-petscarch' and removed
> dependencies for PETSC_ARCH/PETSC_DIR
>
> However configure is still running it after programs.py and few other
> things. [and I don't see any dependencies on it.
>
> Matt, How do I specify that arch.py/petscdir.py should be run at the
> earliest?
>

I have no mechanism for specifying an absolute order. I calculate the
partial order by
topological sort and then an arbitrary total order based order in the
Framework.

It sounds like what we should really do is have a configure module that
sets up logging,
and that modules who want to use logging depend on. Then petscdir would
make that
module depend on itself.

   Matt


> thanks,
> Satish
>



-- 
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/20131029/a0caec4d/attachment.html>


More information about the petsc-dev mailing list