Firedrake Project Conda Recipes

This repository contains a set of (work-in-progress) conda recipes for The Firedrake Project.

The goal is to create a set of conda packages that allow the easy installation of a fully-functioning build of the Firedrake PDE solver.

Current status

Recipes for building PyOP2 and its dependencies have been created.

It is currently possible to build and install these packages on linux-64, but

from pyop2 import op2

does not work:

>>> from pyop2 import op2
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/gmarkall/miniconda/envs/pyop2package/lib/python2.7/site-packages/pyop2/__init__.py", line 9, in <module>
    from op2 import *
  File "/home/gmarkall/miniconda/envs/pyop2package/lib/python2.7/site-packages/pyop2/op2.py", line 39, in <module>
    import base
  File "/home/gmarkall/miniconda/envs/pyop2package/lib/python2.7/site-packages/pyop2/base.py", line 54, in <module>
    from sparsity import build_sparsity
ImportError: libpetsc.so.3.05: cannot open shared object file: No such file or directory

This is likely due to some failure in the process of setting RPATH for some shared object at installation time.

Notes

• Presently PETSc and PyOP2 depend on the mpi4py conda package, which uses mpich2.
• LAPACK 3.5.0 with its embedded BLAS is used at the moment - this is likely to
  be slow. It should be possible to use ATLAS or possibly MKL, but this basic
  setup is used until correct functionality is obtained.
• Recipes for PETSc and PETSc4py point to a local tarball at the moment, due to
  the slowness of the clone from Bitbucket. This could be remedied if there is a
  way to perform a shallow clone for a conda build.
• Building and testing is presently done on a Debian 7.6.0 machine, but moving
  to using Binstar Build would be desirable.
• All building/testing is being done with Python 2.7 initially.