Problem with petscvariables

Barry Smith bsmith at mcs.anl.gov
Tue Dec 18 17:20:41 CST 2007


    The original "plan" was that you either left it in the same place  
or ran config/configure.py
with an appropriate --prefix.  This "hardwiring" I think is mostly for  
the dynamic libraries
since directory names get compiled directly into the .o files. I added  
some support for
using the environmental variables PETSC_DIR and PETSC_ARCH to find the  
appropriate
dynamic libraries (I think this works).

    To be able to move roots around I think we need
1) petscconf gets printed with PETSC_DIR and _ARCH with no hardwiring
2) PETSC_DIR and PETSC_ARCH environomental variables should overwrite
     the hardwired ones when loading dynamic libraries (I think this  
already works).

   So sounds relatively easily doable,

    Barry




On Dec 18, 2007, at 4:58 PM, Matthew Knepley wrote:

> I just built a brand new petsc-dev, but $PETSC_DIR was hardcoded
> everywhere in petscvariables, so when I moved the root directory
> it all broke. Should we define something at the top, and the redirect
> to it everytime?
>
>  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
>




More information about the petsc-dev mailing list