Open tueda opened 6 years ago
Does it give this crash also on the LINUX machines of Nikhef that have an older version of the libraries?
Jos
On 19 okt. 2017, at 15:20, Takahiro Ueda notifications@github.com wrote:
Currently I have a test failure for ParFORM with the ExtComm_1 test on my PC. The strange thing is that it fails via Ruby:
./check.rb ../sources/parform ExtComm_1 (aPLUSbTO2 and aPLUSbTO3 get zero or timeout), while when I run ParFORM directly it works well:
mpirun -np 4 ../sources/parform -D TEST=ExtComm_1 examples.frm For now I guess this might be library version dependent and due to a kind of interference among forking in Ruby, MPI and FORM, maybe. The following is the list of versions of programs/libraries on my PC:
ruby 2.4.2p198 (2017-09-14 revision 59899) [x86_64-linux] gcc (Homebrew gcc 5.4.0) 5.4.0 mpirun (Open MPI) 2.1.2 ParFORM 4.2.0 (Jul 16 2017, v4.2.0-18-gd15ec75) 64-bits — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/vermaseren/form/issues/235, or mute the thread https://github.com/notifications/unsubscribe-auth/AFLxEvv8VnIpqRbTaIb2fh-mA3t6G6hBks5st0ydgaJpZM4P_NzJ.
For example, I have
ruby 2.3.1p112 (2016-04-26 revision 54768) [x86_64-linux]
gcc (Homebrew gcc 5.3.0 --without-glibc) 5.3.0
mpirun (Open MPI) 1.10.2
ParFORM 4.2.0 (Jul 16 2017, v4.2.0-18-gd15ec75) 64-bits
on lippe and it works both directly and via Ruby.
I'm afraid that via Ruby on my PC, the system (ridiculously too old) glibc is somehow linked during the execution of ParFORM and causes such problems...
Currently I have a test failure for ParFORM with the
ExtComm_1
test on my PC. The strange thing is that it fails via Ruby:(
aPLUSbTO2
andaPLUSbTO3
get zero or timeout), while when I run ParFORM directly it works well:For now I guess this might be library version dependent and due to a kind of interference among forking in Ruby, MPI and FORM, maybe. The following is the list of versions of programs/libraries on my PC: