reenable pytest -p unraisableexception and -p threadexception #1898
Conversation
graingert
force-pushed
the
reenable-unraisable-thread-except-hooks
branch
from
04b2f56
to
624736d
Compare
|
|
||
| @contextmanager | ||
| def disable_threading_excepthook(): | ||
| threading.excepthook, prev = _noop, threading.excepthook |
There was a problem hiding this comment.
sadly threading.__excepthook__ is only in 3.10, so I disable the hook rather than restore the old one
There was a problem hiding this comment.
I don't know... if we were already testing 3.10 then maybe it would be good practice to put an if in here to use threading.__excepthook__ when possible and then when we dropped 3.9 there would be a coverage loss and we'd remove the work around. But we aren't testing against 3.10... so... a comment? Or perhaps this is too far forward looking to be appropriate.
There was a problem hiding this comment.
Probably shouldn't hold this up. Maybe I'll add something in #1921.
Codecov Report
@@ Coverage Diff @@
## master #1898 +/- ##
==========================================
- Coverage 99.59% 93.80% -5.80%
==========================================
Files 114 114
Lines 14574 14576 +2
Branches 1110 1113 +3
==========================================
- Hits 14515 13673 -842
- Misses 42 822 +780
- Partials 17 81 +64
|
graingert
force-pushed
the
reenable-unraisable-thread-except-hooks
branch
from
c03e3a9
to
7b65819
Compare
graingert
force-pushed
the
reenable-unraisable-thread-except-hooks
branch
from
e565569
to
6273b1e
Compare
|
See https://gitter.im/python-trio/general?at=602b8e1293539e23437b2daa for the Fedora failure |
| faulthandler_timeout = 60 | ||
| markers = ["redistributors_should_skip: tests that should be skipped by downstream redistributors"] | ||
| junit_family = "xunit2" | ||
| filterwarnings = ["error"] |
There was a problem hiding this comment.
The -W error in ci.sh wouldn't be needed anymore, right?
There was a problem hiding this comment.
I'm pretty sure it's redundant now, but does no harm there
There was a problem hiding this comment.
It'll be confusing when someone wants it off and changes one spot and it's still active. Or when they are looking to Trio as an example.
| @@ -1,9 +0,0 @@ | |||
| [tool:pytest] | |||
There was a problem hiding this comment.
I would prefer that unrelated restructures like this be handled separately. Just commenting, not part of the requested changes.
|
|
||
| @contextmanager | ||
| def disable_threading_excepthook(): | ||
| threading.excepthook, prev = _noop, threading.excepthook |
There was a problem hiding this comment.
I don't know... if we were already testing 3.10 then maybe it would be good practice to put an if in here to use threading.__excepthook__ when possible and then when we dropped 3.9 there would be a coverage loss and we'd remove the work around. But we aren't testing against 3.10... so... a comment? Or perhaps this is too far forward looking to be appropriate.
Co-authored-by: Kyle Altendorf <[email protected]>
| faulthandler_timeout = 60 | ||
| markers = ["redistributors_should_skip: tests that should be skipped by downstream redistributors"] | ||
| junit_family = "xunit2" | ||
| filterwarnings = ["error"] |
There was a problem hiding this comment.
It'll be confusing when someone wants it off and changes one spot and it's still active. Or when they are looking to Trio as an example.
|
|
||
| @contextmanager | ||
| def disable_threading_excepthook(): | ||
| threading.excepthook, prev = _noop, threading.excepthook |
There was a problem hiding this comment.
Probably shouldn't hold this up. Maybe I'll add something in #1921.
by restoring the default unraisablehook, and disabling the threading.excepthook in tests that need it