Jan Beich Reporter
	Description • 10 years ago

The file builds fine with gcc-4.9 but fails with clang 3.5.

In file included from media/webrtc/trunk/webrtc/modules/modules_audio_coding_module/Unified_cpp__audio_coding_module0.cpp:119:
media/webrtc/trunk/webrtc/modules/audio_coding/main/acm2/audio_coding_module_impl.cc:2075:10: error:
      reinterpret_cast from 'nullptr_t' to 'const webrtc::AudioEncoder *' is not
      allowed
  return reinterpret_cast<const AudioEncoder*>(NULL);
         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
media/webrtc/trunk/webrtc/modules/audio_coding/main/acm2/audio_coding_module_impl.cc:2094:10: error:
      reinterpret_cast from 'nullptr_t' to 'const webrtc::ReceiverInfo *' is not
      allowed
  return reinterpret_cast<const ReceiverInfo*>(NULL);
         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2 errors generated.

	Jan Beich Reporter
	Comment 1 • 10 years ago

FreeBSD implements NULL as nullptr in C++11 mode but GCC provides its own NULL via fixincludes. So, rather than casting just use nullptr because it is "implicitly convertible and comparable to any pointer type or pointer-to-member type."

	Randell Jesup [:jesup] (needinfo me)
	Comment 2 • 10 years ago

Comment on attachment 8550835 [details] [diff] [review]
use nullptr

Review of attachment 8550835 [details] [diff] [review]:
-----------------------------------------------------------------

Upstream does not use nullptr, so we really don't want to introduce its usage here.  Some other workaround will be needed.  Worst-case - an ifdef, but I'd prefer to avoid it and instead use some type of casting as needed, or a temporary the compiler will do away with.

	Gian-Carlo Pascutto [:gcp] Assignee
	Comment 3 • 10 years ago

How about just replacing the whole cast by return NULL? Works fine for GCC. What about clang?

	Gian-Carlo Pascutto [:gcp] Assignee
	Comment 4 • 10 years ago

Seems to work with GCC 4.9.2 and clang 3.5 for me.

	Jan Beich Reporter
	Comment 5 • 10 years ago

Comment on attachment 8567857 [details] [diff] [review]
Just return a NULL ptr instead of casting NULL

Works for me as well.

	Gian-Carlo Pascutto [:gcp] Assignee
	Comment 6 • 10 years ago

https://tbpl.mozilla.org/?tree=Try&rev=ae25ca06ae31

	Gian-Carlo Pascutto [:gcp] Assignee
	Comment 7 • 10 years ago

https://hg.mozilla.org/integration/mozilla-inbound/rev/b7bceb058d33

	Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)
	Comment 8 • 10 years ago

https://hg.mozilla.org/mozilla-central/rev/b7bceb058d33

	Jan Beich Reporter
	Comment 9 • 10 years ago

Comment on attachment 8567857 [details] [diff] [review]
Just return a NULL ptr instead of casting NULL

The fix didn't make it in time before the uplift:
http://buildbot.rhaalovely.net/builders/mozilla-central-freebsd-amd64
http://buildbot.rhaalovely.net/builders/mozilla-aurora-freebsd-amd64

Approval Request Comment
[Feature/regressing bug #]: bug 1109248 regression
[User impact if declined]: broken build on FreeBSD or any other platform with #define NULL nullptr in system headers; GCC builds aren't affected due to compiler's hacks
[Describe test coverage new/current, TreeHerder]: Try in comment 6, landed in m-c
[Risks and why]: Low, can only break build.
[String/UUID change made/needed]: None

	Lukas Blakk [:lsblakk] use ?needinfo
	Comment 10 • 10 years ago

Comment on attachment 8567857 [details] [diff] [review]
Just return a NULL ptr instead of casting NULL

low risk, approving.

	Ryan VanderMeulen [:RyanVM]
	Comment 11 • 10 years ago

https://hg.mozilla.org/releases/mozilla-aurora/rev/256f560409ed