[Octave-bug-tracker] [bug #40232] Memory leak in string_vector::resize or load_path::get_file_list

[John W. Eaton]

Follow-up Comment #1, bug #40232 (project octave):

It looks to me like this leak is just happening because the load_path
singleton is not explicitly cleaned up when Octave exits.

If I uncomment the call to singleton_cleanup_list::cleanup in
do_octave_atexit, the valgrind summary for starting and exiting Octave (with
--no-gui) goes from


==15114== LEAK SUMMARY:
==15114==    definitely lost: 592 bytes in 74 blocks
==15114==    indirectly lost: 0 bytes in 0 blocks
==15114==      possibly lost: 218,848 bytes in 4,598 blocks
==15114==    still reachable: 2,092,181 bytes in 7,037 blocks
==15114==         suppressed: 0 bytes in 0 blocks


to


==8620== LEAK SUMMARY:
==8620==    definitely lost: 0 bytes in 0 blocks
==8620==    indirectly lost: 0 bytes in 0 blocks
==8620==      possibly lost: 5,636 bytes in 87 blocks
==8620==    still reachable: 288,992 bytes in 4,355 blocks
==8620==         suppressed: 0 bytes in 0 blocks


As far as I can tell, all the "definitely lost" bytes are due to load_path.

I'd uncomment the singleton cleanup but there is an ominous warning just
before that line that tells me I shouldn't.

I'm not so worried about these kinds of "leaks", though ultimately I'd like to
eliminate them by eliminating the singletons (I have some ideas about how to
do that, but it will take time).

Perhaps we should just close this report?


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?40232>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/