844869 - NS_ASSERTION should be fatal with GTest

Status: RESOLVED FIXED

(Testing :: General, defect)

Comment 1

[Benoit Girard (:BenWa)]

Looks like we need XPCOM_DEBUG_BREAK=stack-and-abort:

http://mxr.mozilla.org/mozilla-central/source/xpcom/base/nsDebugImpl.cpp#211

Comment 2

[Benoit Girard (:BenWa)]

Attachment: patch

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Comment on attachment 717911 [details] [diff] [review]
patch

Are you sure you don't want the final parameter to be true?

It seems a little cleaner doing this in whatever (make/python/shell) code runs this, but I suppose doing it in the C++ is fine.

Please test that an NS_ASSERTION firing actually causes a failure.

Comment 4

[Benoit Girard (:BenWa)]

(In reply to David Baron [:dbaron] (don't cc:, use needinfo? instead) from comment #3)
> Comment on attachment 717911 [details] [diff] [review]
> patch
> 
> Are you sure you don't want the final parameter to be true?
> 

Well I though this would be the default value. If something wanted to override the default I think it make sense to let it. If you have an example where this would be bad I can change it.

> It seems a little cleaner doing this in whatever (make/python/shell) code
> runs this, but I suppose doing it in the C++ is fine.

Part of my motivation is I'd like ./firefox -unittest to give you a sane default run of this. This makes it easy to run in a debugger for example.

> 
> Please test that an NS_ASSERTION firing actually causes a failure.

Done

Comment 5

[Benoit Girard (:BenWa)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/00fc2c679bf5

Comment 6

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/00fc2c679bf5