[Development] Fw: Change in qt/qt5[dev]: Updated submodules.
Kai.Koehne at theqtcompany.com
Thu Nov 26 10:10:50 CET 2015
> -----Original Message-----
> From: Development [mailto:development-bounces at qt-project.org] On Behalf
> Of Marc Mutz
> Sent: Thursday, November 26, 2015 9:53 AM
> To: development at qt-project.org
> Subject: Re: [Development] Fw: Change in qt/qt5[dev]: Updated submodules.
> On Wednesday 25 November 2015 20:32:50 Thiago Macieira wrote:
> > On Wednesday 25 November 2015 18:27:11 Aleix Pol wrote:
> > > > Anyway, if this is a compiler bug, the conditions for triggering
> > > > it may not be evident.
> > >
> > > Does the fact that it's a compiler bug change the fact that it
> > > should be changed on the Qt side of things?
> > >
> > > If we have this compiler on the CI, Qt users will have it as well.
> > No, it does not change the fact that we need to either work around or
> > remove the change that triggered it. Qt needs to compile with MSVC 2015.
> > One way to do that is to disable noexcept for MSVC2015. Another is to
> > remove the use of <type_traits>.
> We cannot remove the <type_traits> use, because we cannot replace it with the
> noexcept operator we used before, because that already failed. We cannot
> remove noexcept in general, because it's now part of the API. We can, however,
> remove it for just MSVC.
> > If the change is to remove noexcept, I'd like to know if Update 1
> > fixes the issue. If it doesn't, it would be a good idea to dissect a
> > testcase and submit to Microsoft. It would be the fourth one for
> > Update 1 that we found just by compiling Qt with all the C++11 features
> Here's a test case for what I think might be the problem:
The test in this and the following patch set compile fine for me with MSVC 2015 (32 bit),
while qtenginio fails. So it's not actually testing the problem.
More information about the Development