623126 - Add constructor for nsDebugImpl, nsTraceRefcntImpl, EmptyEnumeratorImpl, and nsSimpleUnicharStreamFactory to placate CLang

Status: RESOLVED FIXED

(Core :: XPCOM, defect)

Description

[Rafael Ávila de Espíndola (:espindola) (not reading bugmail)]

Attachment: patch

nsSimpleUnicharStreamFactory is missing a user defined constructor,
but in nsUnicharInputStream.cpp a const variable of this type is defined.

This is not valid c++. For more information see "Default initialization of
const variable of a class type requires user-defined default constructor" in
http://clang.llvm.org/compatibility.html#c++

Comment 1

[timeless]

The reporter's summary and initial comment were both lame. I'm merely adjusting
the summary and providing a better link. I am not passing judgement on the
quality of the bug report.

http://clang.llvm.org/compatibility.html#default_init_const

Comment 2

[Benjamin Smedberg]

Just as with the other bug, the missing constructor is intentional so that GCC does not emit a initialization function. Absent a detailed explanation from a language lawyer, I don't think I want to accept this change.

Comment 3

[Rafael Ávila de Espíndola (:espindola) (not reading bugmail)]

Given the discussion on bug 623123, are you ok with it?

Comment 5

[(no longer active)]

http://hg.mozilla.org/mozilla-central/rev/ca41c5663999

Comment 7

[Jonathan Wakely]

(In reply to comment #2)
> Just as with the other bug, the missing constructor is intentional so that GCC
> does not emit a initialization function. Absent a detailed explanation from a
> language lawyer, I don't think I want to accept this change.

A bit late, but language lawyer here, see http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44499#c2 for chapter and verse. As pointed out by some of the dups of this bug, GCC 4.6 rejects this too, see the note I added to http://gcc.gnu.org/gcc-4.6/changes.html#cplusplus

Comment 8

[Jonathan Wakely]

I see that the commit adds empty user-defined constructors. Did anyone consider the alternative of adding an initializer instead, as I suggested in the GCC 4.6 changes and in Bug 645469 c5?
That might avoid emitting an initialization function.

Comment 9

[Mike Hommey [:glandium]]

(In reply to comment #8)
> I see that the commit adds empty user-defined constructors. Did anyone consider
> the alternative of adding an initializer instead, as I suggested in the GCC 4.6
> changes and in Bug 645469 c5?
> That might avoid emitting an initialization function.

I don't think it makes any difference in the generated code.

Comment 10

[Alex Vincent [:WeirdAl]]

requesting approval for mozilla-2.0 branch:  with Fedora 15, gcc 4.6.0, FF4 will not compile.  #developers pointed me to this bug.

Comment 11

[Benjamin Smedberg]

the 2.0 branch doesn't have any planned updates, what is the purpose of the nomination?

Comment 12

[[e-mail]]

I wonder why this is marked as resolved fix?  In Fedora 15 it is still a problem, was the patch applied upstream and then not brought into Fedora?  I'm trying to rebuild sunbird, and this bug in xulrunner is breaking the sunbird build.  I just created an updated version of the xulrunner rpm using this patch to fix the problem, but if there is something upstream I can grab it is probably better to use that.