Closed klauer closed 1 year ago
Well it failed spectacularly on GitHub Actions, whereas it passed locally so 🤷 Going to mark profiling code as xfail temporarily (PR is still draft)
OK, this is a bit of a mess and there are a ton of changes. I did this over a long period of time with (sometimes) limited attention being paid to what I was doing.
Each change should be really scrutinized!
I notice that you left some of the cleanup for another day- I think that's completely appropriate. You've cleaned up almost all of it already and the last QMenus seem to be lurking from upstream code.
🤞
2 green checkmarks in a row indicate it's probably good enough as-is. Before this gets any larger, if you guys are good with it - I'd say let's merge @tangkong @ZLLentz
We may revisit and adjust how strict this is in the future based on how many intermittent failures we get (due to gc/teardown slowness). Increasing the wait time may also suffice...
I'll click merge now to make sure it happens
Description
ophyd.Kind
enum.Enum
, being anIntFlag
is a bit of a special beast. Its functionality in Python 3.11 differs in enumeration of items - resulting in typhos only picking up a couple of component kinds.multiprocessing
to run a caproto IOC ended up being problematic for Python 3.11, so I restructured the test suite a bit to run IOCs by name (simply regenerating the test cases in the parent process and in the IOC)load_ui_file
and modified it so we can always get ourDisplay
instance back and clean it up if desirable.SignalConnection.remove_connection
Motivation and Context
Python 3.11 support would be nice eventually
This PR manifested itself like this:
multiprocessing
and isolate the test IOC in its own scripttest_take_top_level_widget_screenshots
sees hundreds of QMenu instances at the top level)Overall, I think this is in a better spot than before. It's absolutely not perfect, though.
How Has This Been Tested?
Locally and test suite
Where Has This Been Documented?
Screenshots (if appropriate):