[Development] Removing the -c++11 option from configure

Thiago Macieira thiago.macieira at intel.com
Tue Mar 24 19:53:32 CET 2015


On Tuesday 24 March 2015 19:40:32 Stephen Kelly wrote:
> Thiago Macieira wrote:
> > The reason we prefer -std=c++11 is that it allows us to write code that
> > doesn't trip GNU extensions unlikely to be found on MSVC and other
> > compilers.
> 
> What is the likelyhood of that? No one is going to accidentally add an
> `__attribute ((strong))` namespace in Qt.

Why should we tempt fate?

> In particular, what features are you afraid of accidentally using (locally
> on your machine before CI gets a look) which are available with -std=gnu++11
> but not available with -std=c++11?

I'd like people to catch that before submitting to the CI. That's why the new 
headersclean is coming in, for example. That's why -Werror is enabled by 
default.

You're welcome to turn any of those checks off (configure options, editing the 
mkspec, etc.), but be prepared to get yelled at if by doing that you miss 
something.

> > User code has been using this flag just fine for a couple of years, so I
> > don't think they need -std=gnu++11.
> 
> Yes. I did see a lot of issues pop up around mingw, but as you say, Qt
> people haven't noticed anything. Is that really a good reason not to fix it?
> Unreported bugs are fixed all the time.

Fix what? Until someone says this is a problem, we don't know if it is.

There's a difference between "unreported bug" and "not a bug".

-- 
Thiago Macieira - thiago.macieira (AT) intel.com
  Software Architect - Intel Open Source Technology Center




More information about the Development mailing list