Repository navigation
fs: use try...finally for abort race - #67545
Conversation
Signed-off-by: Renegade334 <rene@renegade334.me.uk>
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #67545 +/- ##
==========================================
- Coverage 92.71% 90.68% -2.04%
==========================================
Files 430 800 +370
Lines 196040 280935 +84895
Branches 30067 54281 +24214
==========================================
+ Hits 181757 254754 +72997
- Misses 13956 16263 +2307
- Partials 327 9918 +9591
🚀 New features to boost your workflow:
|
LiviaMedeiros
left a comment
There was a problem hiding this comment.
IMHO this should be dont-land-on-v27.x+, i.e. opened against previous release lines while main can move forward using using.
LGTM if the plan is to land this as is (backportable) and immediately revert (non-backportable) on main.
This is one of those cases where there simply isn't any practical advantage beyond code brevity. It doesn't change the behaviour at all here. Keeping PRs landable across multiple release lines is a much greater advantage. Agree that ultimately we can migrate all of these cases, but there are dozens across the codebase doing no harm to be left alone for now. |
|
Landed in c91d8aa |
As per #64230.
try...finallyprovides identical semantics here (the disposer fromaddAbortListener()does not throw) but avoids the dependence on the flagged language feature in release branches prior to v27.x.