776871 - Enable the use of nullptr for vc10 builds or later

Status: RESOLVED FIXED

(Core :: XPCOM, defect)

Comment 1

[Jim Mathies [:jimm]]

Attachment: patch

Seems like the right way to do this, not sure though...

Comment 2

[Aryeh Gregor (:ayg) (no longer with Mozilla)]

Why is this needed at all?  Bug 626472 adds a configure-time check to configure.in, which defines HAVE_NULLPTR using AC_TRY_COMPILE, i.e., if "int* foo = nullptr;" compiles successfully.  Does this check fail in VC versions that actually support nullptr?  If so, why?

Comment 3

[Jim Mathies [:jimm]]

I'll take a look. HAVE_NULLPTR isn't indexed into mxr yet on mc.

Comment 4

[Jim Mathies [:jimm]]

So $COMPILE_ENVIRONMENT is set to 1 on windows, and this sets SKIP_COMPILER_CHECKS=1, which causes configure to skip over the nullptr check.

Comment 5

[Jim Mathies [:jimm]]

(In reply to Jim Mathies [:jimm] from comment #4)
> So $COMPILE_ENVIRONMENT is set to 1 on windows, and this sets
> SKIP_COMPILER_CHECKS=1, which causes configure to skip over the nullptr
> check.

Hmm, so it's not COMPILE_ENVIRONMENT, SKIP_COMPILER_CHECKS is set to 1 farther up..

Comment 6

[Jim Mathies [:jimm]]

here it is - 

http://mxr.mozilla.org/mozilla-central/source/configure.in#739

Comment 7

[Jim Mathies [:jimm]]

Attachment: patch

will seek reviews on these once I get through a full build.

Comment 8

[Aryeh Gregor (:ayg) (no longer with Mozilla)]

So it looks like Windows skips all compiler checks due to bug 58981, which landed in . . . 2002.  Could we try not disabling compiler checks for Windows, maybe?

Comment 9

[Jim Mathies [:jimm]]

(In reply to :Aryeh Gregor from comment #8)
> So it looks like Windows skips all compiler checks due to bug 58981, which
> landed in . . . 2002.  Could we try not disabling compiler checks for
> Windows, maybe?

I'd rather we file a follow up on that.

Comment 10

[Benjamin Smedberg]

No, we cannot do the compiler checks on Windows without completely rewritting AC_TRY_COMPILE which assumes a GCC-style compiler syntax.

Comment 11

[Masatoshi Kimura [:emk]]

msvc10 is supposed to support nullptr for native codes.
http://msdn.microsoft.com/ja-jp/library/4ex65770%28v=vs.110%29.aspx
http://msdn.microsoft.com/ja-jp/library/4ex65770%28v=vs.100%29.aspx
http://msdn.microsoft.com/ja-jp/library/4ex65770%28v=vs.90%29.aspx
Why nullptr is enabled only for msvc11?

Comment 12

[Jim Mathies [:jimm]]

(In reply to Masatoshi Kimura [:emk] from comment #11)
> msvc10 is supposed to support nullptr for native codes.
> http://msdn.microsoft.com/ja-jp/library/4ex65770%28v=vs.110%29.aspx
> http://msdn.microsoft.com/ja-jp/library/4ex65770%28v=vs.100%29.aspx
> http://msdn.microsoft.com/ja-jp/library/4ex65770%28v=vs.90%29.aspx
> Why nullptr is enabled only for msvc11?

I had no idea. I'll fire up a build with vc10 to test.

Comment 13

[Jim Mathies [:jimm]]

(In reply to Jim Mathies [:jimm] from comment #12)
> (In reply to Masatoshi Kimura [:emk] from comment #11)
> > msvc10 is supposed to support nullptr for native codes.
> > http://msdn.microsoft.com/ja-jp/library/4ex65770%28v=vs.110%29.aspx
> > http://msdn.microsoft.com/ja-jp/library/4ex65770%28v=vs.100%29.aspx
> > http://msdn.microsoft.com/ja-jp/library/4ex65770%28v=vs.90%29.aspx
> > Why nullptr is enabled only for msvc11?
> 
> I had no idea. I'll fire up a build with vc10 to test.

Appears to be building initially, if this makes it through I'll update the version to 1600.

Comment 14

[(no longer active)]

(In reply to Jim Mathies [:jimm] from comment #13)
> (In reply to Jim Mathies [:jimm] from comment #12)
> > (In reply to Masatoshi Kimura [:emk] from comment #11)
> > > msvc10 is supposed to support nullptr for native codes.
> > > http://msdn.microsoft.com/ja-jp/library/4ex65770%28v=vs.110%29.aspx
> > > http://msdn.microsoft.com/ja-jp/library/4ex65770%28v=vs.100%29.aspx
> > > http://msdn.microsoft.com/ja-jp/library/4ex65770%28v=vs.90%29.aspx
> > > Why nullptr is enabled only for msvc11?
> > 
> > I had no idea. I'll fire up a build with vc10 to test.
> 
> Appears to be building initially, if this makes it through I'll update the
> version to 1600.

Please make sure to make a preprocessed version of a source file (by doing make nsFoo.i in the corresponding objdir subdirectory for nsFoo.cpp) and check to see if nullptr is indeed being used in there.

Comment 15

[(no longer active)]

Comment on attachment 645267 [details] [diff] [review]
patch

Clearing the review for now.

Comment 16

[Jim Mathies [:jimm]]

Attachment: patch

Looks like this is giving expected output, but to be sure - generating a test.i for a test.cpp that contains:

static void test()
{
  char * p = nsnull;
  char * d = nullptr;
}

I get:

(a whole bunch of white space and comments plus..)

static void test()

{

  char * p = nsnull;

  char * d = nullptr;

}

Comment 17

[Jim Mathies [:jimm]]

Or maybe this is the test you're looking for -

test.cpp:

static void test()
{
#ifndef HAVE_NULLPTR
  char * p = nsnull;
#else
  char * d = nullptr;
#endif
}


test.i:

static void test()
{

  char * d = nullptr;

#line 10 "f:/Mozilla/firefox/mc/widget/windows/test.cpp"

}

Comment 18

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/8b5b175234df