Conversation
this way we actually saw the error
Codecov Report
@@ Coverage Diff @@
## main #54 +/- ##
==========================================
+ Coverage 70.12% 70.83% +0.71%
==========================================
Files 5 5
Lines 164 168 +4
==========================================
+ Hits 115 119 +4
Misses 49 49
Flags with carried forward coverage won't be shown. Click here to find out more.
|
sloede
left a comment
sloede
left a comment
There was a problem hiding this comment.
Thanks a lot for providing this hot fix! Maybe you can also create an issue that lets us track this such that we cannot forget to fix it once there is a canonical solution available from MPI.jl.
| // This would better be fixed using some functionality of MPI.jl | ||
| const char * call_MPI_run_init_hook = | ||
| "using MPI;\n" | ||
| "if MPI.Initialized() && MPI.JULIA_TYPE_PTR_ATTR[]==0;\n" |
There was a problem hiding this comment.
Since we are already tinkering with MPI.jl's internals here, why not use the internal variable that is set in run_init_hooks,
to check if we need to run it again? That is, something like
"if MPI.Initialized() && !MPI._finished_loading[];\n"I am just concerned that this check might fail if JULIA_TYPE_PTR_ATTR is left == 0 by some MPI implementation.
2 participants
Hotfix, that hopefully can be removed again later.
JuliaParallel/MPI.jl#746