Bug 211089 – [JSC] Add $vm.assertEnabled() to suppress Debug crash expected tests in release+assert build

• Status: RESOLVED
• Resolution: FIXED
• Component: JavaScriptCore

[Yusuke Suzuki]

We have ASSERT which only fires when DOM object is wrongly implemented to catch wrong ones.
We are using ASSERT in JSC side since we don't know all the DOM objects are correctly implemented. If it is not correctly implemented, in Release build, JSC is returning sub-optimal results. But in Debug build, we can catch the wrong ones, which allows us to fix the wrong thing while avoiding catastrophic crashes in release build.

However, the test is assuming that ASSERT is not enabled if it is not Debug build. Release & Debug builds work, but Release+Assert does not meet this assumption.

So, skip this based on `$vm.assertEnabled()`.

[Yusuke Suzuki]

Created attachment 397713 [details]
Patch

[Yusuke Suzuki]

Created attachment 397715 [details]
Patch

[Keith Miller]

Comment on attachment 397715 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=397715&action=review

r=me with suggestion

> JSTests/stress/incorrect-put-could-generate-invalid-ic-involving-dictionary-flatten.js:19
> +if (!$vm.assertEnabled()) {

Why not `if (!$vm.assertEnabled()) quit()`

[Yusuke Suzuki]

Comment on attachment 397715 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=397715&action=review

>> JSTests/stress/incorrect-put-could-generate-invalid-ic-involving-dictionary-flatten.js:19
>> +if (!$vm.assertEnabled()) {
> 
> Why not `if (!$vm.assertEnabled()) quit()`

Sounds good!

[Yusuke Suzuki]

Committed r260784: <https://trac.webkit.org/changeset/260784>

[Radar WebKit Bug Importer]

<rdar://problem/62468324>