We declare PVFS_sys_release and PVFS_mgmt_release functions in our external headers. Is there a good reason for us doing that? My concern is that those release functions just do cleanup of the state machine pointers, but don't cancel any associated jobs that may have been posted. Given that we don't actually use the operation ids to test for completed jobs (the test and testsome just end up calling test_context), allowing users to call PVFS_sys_release could cause nasty corruption.
pvfs2-client-core calls sys_release directly, but it calls other internal (PINT_*) functions as well, so making the release functions internal seems ok.
Incidentally, we don't actually provide a PVFS_sys_cancel, that does true cancellation of an isys request. Should we?
-sam _______________________________________________ Pvfs2-developers mailing list [email protected] http://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers
