Avoid pointless discards during the reuse and target phases
#3862
+39
−9
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Zac-HD
added
performance
5 tasks
|
Nice! Probably closes #3446, too. |
Zac-HD
force-pushed
the
uncap-replay-buffers
branch
from
5292d4b to
7700e75
Compare
Zac-HD
force-pushed
the
uncap-replay-buffers
branch
2 times, most recently
from
d2323e5 to
91c63bb
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This problem was jointly identified in discussion of #3827 with @tybug and @hgoldstein95. Finding this kind of performance issue without knowing to look for it is a great illustration of the value of observability!
It turns out that we're spuriously aborting many test cases with status=Overrun, due to
ConjectureData.from_buffer()limiting the max length to the length of the provided bytestring unlessextend=...is passed. This makes sense in theshrinkphase - any longer buffer cannot be a successful shrink, so we may as well abort early - but not in thereuseortargetphases.reusephase is easy to fix: simply add the relevant argument to allow extending buffers replayed from the database - we'll get different behaviour, but get a performance gain (and maybe avoidHealthCheck.filter_too_much) by finishing this test case rather than retrying in thegeneratephase.targetphase is a little more complicated, and ... it turns out that by the time I worked out how we'd need to thread state through to the shrinker I had 80% of a fix, so this is now a PR rather than an issue 😅Tests tbd when I have some more time.