You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The async-gc helper uses FinalizationRegistry.prototype.cleanupSome. Probably at the time the helper was written (2019), this feature was part of a stage 3 proposal, but it was split off, demoted to stage 2 in the meantime, and stalled. It's probably not desirable to have part of the test harness depend on a stage 2 proposal which may only slowly (if ever) advance to stage 3.
Outside of the tests for FinalizationRegistry.prototype.cleanupSome itself, the helper is used in 3 tests:
It's not clear to me whether any of the tests that use this harness helper are actually regularly executed. test262.fyi results show only passing results for SpiderMonkey, Serenity LibJS, and GraalVM. It could be that any test using this helper is in practice a coverage gap for most implementations due to being skipped, or marked as expected-fail.
I don't know about the official implementation status of cleanupSome, but as far as I can tell from looking for its presence in a recently updated esvu installation:
engine262: feature flag exists but does not define method
JavaScriptCore: no
QuickJS: no
Serenity LibJS: yes
SpiderMonkey: yes
V8: behind flag
XS: yes
The resolution of this issue should be to create a PR removing cleanupSome from the test corpus, since it is stage 2, and seek feedback from implementors on whether they would miss it. Also, investigate whether the functionality being tested in those remaining 3 tests can also be tested without cleanupSome, and whether any changes to the interpreting requirements would be necessary to do so, and include such changes in the PR if possible.
The text was updated successfully, but these errors were encountered:
The async-gc helper uses
FinalizationRegistry.prototype.cleanupSome
. Probably at the time the helper was written (2019), this feature was part of a stage 3 proposal, but it was split off, demoted to stage 2 in the meantime, and stalled. It's probably not desirable to have part of the test harness depend on a stage 2 proposal which may only slowly (if ever) advance to stage 3.Outside of the tests for
FinalizationRegistry.prototype.cleanupSome
itself, the helper is used in 3 tests:It's not clear to me whether any of the tests that use this harness helper are actually regularly executed. test262.fyi results show only passing results for SpiderMonkey, Serenity LibJS, and GraalVM. It could be that any test using this helper is in practice a coverage gap for most implementations due to being skipped, or marked as expected-fail.
I don't know about the official implementation status of cleanupSome, but as far as I can tell from looking for its presence in a recently updated esvu installation:
The resolution of this issue should be to create a PR removing
cleanupSome
from the test corpus, since it is stage 2, and seek feedback from implementors on whether they would miss it. Also, investigate whether the functionality being tested in those remaining 3 tests can also be tested withoutcleanupSome
, and whether any changes to the interpreting requirements would be necessary to do so, and include such changes in the PR if possible.The text was updated successfully, but these errors were encountered: