allow user to configure opt_handle_signals #333
Conversation
Codecov Report
@@ Coverage Diff @@
## main #333 +/- ##
==========================================
- Coverage 41.50% 41.50% -0.01%
==========================================
Files 76 76
Lines 4652 4653 +1
==========================================
Hits 1931 1931
- Misses 2721 2722 +1
|
|
Thank you. Could you also update the FAQ.md please with better advice for multithreading? |
brian-dellabetta
force-pushed
the
feature/bd/handle-opt-signal-choice
branch
from
6385f91
to
e9b7d18
Compare
|
@cjdoris sure thing, just pushed and referenced the discussion thread. |
| @@ -129,7 +129,7 @@ def args_from_config(): | |||
| CONFIG['opt_sysimage'] = sysimg = path_option('sysimage', check_exists=True)[0] | |||
| CONFIG['opt_threads'] = int_option('threads', accept_auto=True)[0] | |||
| CONFIG['opt_warn_overwrite'] = choice('warn_overwrite', ['yes', 'no'])[0] | |||
| CONFIG['opt_handle_signals'] = 'no' | |||
| CONFIG['opt_handle_signals'] = choice('handle_signals', ['yes', 'no'], default='no')[0] | |||
There was a problem hiding this comment.
Can we default to yes? Defaulting to a behaviour which only causes crashes for backward compatibility doesn't seem very useful.
There was a problem hiding this comment.
Can we default to
yes? Defaulting to a behaviour which only causes crashes for backward compatibility doesn't seem very useful.
I agree, that's probably better to default to yes, but I left off for now as it is a behavioral change. Just posted a comment on this thread too, let's see what the author thinks
|
Thanks! Please open a separate issue/PR to change the default - AFAIR making it yes by default has undesirable side-effects on Python. |
Hi @cjdoris , this PR is to allow a user to configure what is passed to
julia --handle-signals=<yes,no>. I have just updated it to be in line with all the other CONFIG params, defaulting tonoto retain backwards compatibility.Following discussion thread starting here, I have found that this is much more robust and clean way to allow for multi-threaded julia code than my previous workaround.
Related to #219
Related to #298
Related to #330