Use STDIN project in test_projects to ensure it runs quickly #2575
Conversation
nipunn1313
force-pushed
the
cleanup_test
branch
from
161f0df
to
d3a527a
Compare
Existing test was actually running a full black-primer run which could be slow. This goes from 8 seconds to 0.4 seconds on my machine. Needed to move to top level scope to leverage the caplog feature of pytest in order to test that the command line was parsing the bogus arguments and dumping to stderr.
nipunn1313
force-pushed
the
cleanup_test
branch
from
d3a527a
to
c13b038
Compare
|
You pinged the wrong person fyi :p hopefully they aren't bothered |
ichard26
added
C: maintenance
|
🤦 |
There was a problem hiding this comment.
Still surprisingly slow at ~1.2s on my machine but this is good enough (and won't slow build + test cycles by any reasonable margin). Thank you for the quick turnaround!
FWIW while it's great you're working on black-primer, I'm currently writing a semi-replacement for the development tool: https://github.com/ichard26/diff-shades. Fundamentally it takes the idea of black-primer and extends it with whatever great ideas mypy-primer added (theirs was inspired by ours), notably diffing support between runs. It'll take a long time before diff-shades is at feature and quality parity, but I thought I'd let you know.
Basically for now if there's any ideas you have for primer but that seem too complicated to implement in primer I'd be happy to investigate if it'd be a good fit for diff-shades.
There was a problem hiding this comment.
Speed is good. I did have to take a few passes tho to see how this worked. Like @ichard26 I was amazed that pytest log magic and click's magic get along.
I was too. I actually couldn't get it to work any other way.
Good to know! Thank you. That would be really valuable! I've used mypy-primer and the diff mode is fantastic. If implemented in diff-shades then you can get valuable feedback running it against large projects that don't fully use black yet (eg django!), especially when trying to implement a stability policy. |
Existing test was actually running a full black-primer run which could be slow. This goes from 8 seconds to 0.4 seconds on my machine. Needed to move to top level scope to leverage the caplog feature of pytest in order to test that the command line was parsing the bogus arguments and dumping to stderr.
Fixes @iChard's comment in #2555 (comment)
Runs in 0.4 seconds rather than 8 seconds previously (on my machine).
Checklist - did you ...
(none needed)