702877 - Remove NS_OVERRIDE and replace all uses of it with MOZ_OVERRIDE

Status: RESOLVED FIXED

(Core :: General, defect)

Description

[Jeff Walden [:Waldo]]

NS_OVERRIDE and MOZ_OVERRIDE both indicate the same thing: that the virtual function member being annotated must override a virtual function member on a base class.  MOZ_OVERRIDE abstracts away new C++11 goodness and actually has teeth in some recent compilers.  NS_OVERRIDE is a user attribute and has no teeth except in static analysis builds, and I don't even know what the status of those is any more.

Probably static analysis can look for MOZ_OVERRIDE in new enough gcc/clang compilers (4.7+ and 3.0+, respectively, but note gcc 4.7 isn't released yet), so NS_OVERRIDE and all uses of it should be removed and replaced with MOZ_OVERRIDE.

Comment 1

[Joshua Cranmer [:jcranmer]]

Some notes here:
1. NS_OVERRIDE is generally done before the method declaration, whereas the override keyword would have to be after the method declaration. So all uses of NS_OVERRIDE need to be moved around.

2. It should be possible to alias MOZ_OVERRIDE to the __attribute__((user("NS_override"))) version in sufficiently old (i.e., gcc 4.6-or-less) builds.

Comment 2

[Joshua Cranmer [:jcranmer]]

Attachment: Replace all uses in mozilla-central

This kills all uses of NS_OVERRIDE and deletes the macro to make sure no one else tries to use it again.

Comment 3

[Joshua Cranmer [:jcranmer]]

Attachment: Version that passes tinderbox

Silly me, why did I assume it worked on all platforms just because it worked on my computer?

Comment 5

[Joshua Cranmer [:jcranmer]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/950048a58302

Comment 6

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/950048a58302

Comment 7

[Marco Castelluccio [:marco]]

There's still xpcom/analysis/override.js that refers to NS_OVERRIDE.

Comment 8

[(no longer active)]

(In reply to comment #7)
> There's still xpcom/analysis/override.js that refers to NS_OVERRIDE.

Good catch.  I think that entire script should be removed.  Can you please file a bug on that?  Thanks!