Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: [OMPI devel] MPIT solution still wrong
From: Jeff Squyres (jsquyres) (jsquyres_at_[hidden])
Date: 2013-08-19 21:03:51


It looks like https://svn.open-mpi.org/trac/ompi/changeset/29043 is a stopgap, but it is still definitely wrong.

The MPIT stuff does *not* compile the same way the C bindings compile. Here's how the C bindings compile:

a) in ompi/mpi/c/profile: always compile libmpi_c_pmpi.la
b) in ompi/mpi/c:
  1. always compile libmpi_c.la (a small number of stub functions and global variables)
  2. only compile libmpi_c_mpi.la if weak symbols are not supported (e.g., OSX)

Then libmpi.la always contains a) and b1). If b2) exists, it is also slurped into libmpi.la.

MPIT is currently compiled as:

a) in ompi/mpi/tool/profile: always compile libmpi_pmpit.la
b) in ompi/mpi/tool: always compile libmpi_mpit.la

It really needs to be done the same way the C bindings are done -- split b) into the global constants that you always need and always compile that, and then optionally compile the rest if weak symbols are not supported.

-- 
Jeff Squyres
jsquyres_at_[hidden]
For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/