Conversation
|
This also causes some compile errors: |
|
Reverted in #19423 |
Where did you see the error ? Travis was ok https://travis-ci.org/flutter/flutter/builds/404310972 |
|
In our post-commit testing. Not sure why travis didn't catch it. Here are some others: |
|
These are all devicelab tests (see |
* flutter_master: (810 commits) Revert engine roll to 316b026 roll engine to 316b026 (flutter#19419) Revert "enable lint unnecessary_const (flutter#19342)" (flutter#19423) enable lint unnecessary_const (flutter#19342) Chevrons in month picker are semi-transparent when the month is scrolled (flutter#19363) Revert "Use FlutterProject to locate files (flutter#18913)" (flutter#19409) Extra debug information in run_release_test (flutter#19405) Fix typo (flutter#19402) Use FlutterProject to locate files (flutter#18913) Revert "roll engine to 9af920e (flutter#19365)" (flutter#19376) roll engine to 9af920e (flutter#19365) increase cache size if image is loaded that is larger than max size (flutter#19352) Add Bash and Zsh command-line completion for flutter (flutter#19243) Support keyboardAppearance field for iOS (flutter#19244) Add option to silence driver extension errors (flutter#19247) Add HeroController to CupertinoApp (flutter#19326) have text finder convert Text.rich to plain text for comparison (flutter#19270) Correct contentPadding type in InputDecoration.debugFillProperties (flutter#19318) Revert "Revert "Use runTests in fuchsia tester. (flutter#19178)" (flutter#19321)" (flutter#19327) Roll engine to c5a63d (flutter#19288) ...
|
I think the failures come from tests running in Dart1 (dev/devicelab/bin/tasks/hot_mode_dev_cycle__benchmark_dart1.dart and dev/devicelab/bin/tasks/hot_mode_dev_cycle_ios__benchmark_dart1.dart). AFAIK it's not possible to use optional new/const with dart1 so the only solution seems to remove the tests running in dart1 (and also all the remaining dart1 things) WDYT? |
|
I tried compiling the pub package with and without new (using I also did not observe a meaningful difference in compile time, 0m13.391s vs 0m13.357s. I tried a basic benchmark with millions of allocations and couldn't repro a slowdown there either. |
This reverts commit cc1cf13.
No description provided.