Add array/hash implicit allocation tests #10339
Conversation
test/ruby/test_allocation.rb
Outdated
| empty_array = empty_array = [] | ||
| empty_hash = empty_hash = {} | ||
| array1 = array1 = [1] | ||
| hash1 = hash1 = {a: 2} | ||
| nill = nill = nil | ||
| block = block = lambda{} |
There was a problem hiding this comment.
Not sure why these are assigned to twice.
There was a problem hiding this comment.
To avoid verbose warnings for the calls that do not use them.
The main issue with that approach is that it stops at the first failure. I suppose I could work around that, building an array of all cases that fail, and then report it at the end if the array isn't empty. One area this doesn't currently test is use of ruby2_keywords. I'd like to add tests for that, not only as a method modifier, but more importantly, when passing the flagged hash in a splat to all of the other method types. |
These are designed to prevent allocation regressions (commits that increase the number of implicitly allocated arrays and hashes). We have already had three commits in the last couple weeks to fix allocation regressions: * 15dc3aa * aceee71 * c388784 This test suite should hopefully allow us to find such regressions in CI before commit, to avoid committing future allocation regressions. This uses assert_separately around each set of tests. Doing it for each individual check was too slow. Failures are gathered and reported at the end of the the suite as a single assertion, with the message describing all failures.
jeremyevans
force-pushed
the
implicit-allocation-tests
branch
from
cb41b67
to
06a4ca8
Compare
|
I pushed a rebased commit with a much faster approach, using one |
These are designed to prevent allocation regressions (commits that increase the number of implicitly allocated arrays and hashes). We have already had three commits in the last couple weeks to fix allocation regressions:
This test suite should hopefully allow us to find such regressions in CI before commit, to avoid committing future allocation regressions.
This uses assert_separately to test each case, which is very slow. Hopefully there is a better way that can isolate the entire test suite, instead of isolating each individual check. This does need some isolation as it disables garbage collection to get accurate allocation counts, and it cannot be run concurrently with other threads in the same process.