Re: stdio.h patches for g++ -std=c++11

[zosrothko <zosrothko at orange dot fr>]

Le 09/05/2014 12:03, Richard Earnshaw a écrit :
> On 09/05/14 11:01, zosrothko wrote:
> > Corinna
> > Le 09/05/2014 10:43, Corinna Vinschen a écrit :
> > > On May  9 07:52, zosrothko wrote:
> > > > Hi Corinna
> > > >
> > > > I made a typo in my previous patch. Here the corrected one.
> > > Thanks.  Your patches are missing a matching ChangeLog entry, btw.
> > >
> > > However, on second thought I'm not sure that removing the 'i' variations
> > > of the functions from "__cplusplus >= 201103L" in this way is such a
> > > good idea.
> > >
> > > The 'i' functions are a newlib extension for embedded targets.
> > >
> > > They are not part of any standard, so they are certainly neither part
> > > of "__cplusplus >= 201103L", nor part of "__STDC_VERSION__ >= 199901L".
> > >
> > > So, either we remove these functions from both standards, or we include
> > > them in both standards.
> > >
> > > But if we remove them from both standards when will they be defined at
> > > all?  This should be cleared up before making this change.
> > >
> > > Is defining them with only
> > >
> > >     #if !defined(__STRICT_ANSI__)
> > >
> > > sufficient?
> > That won't work because of this
> >
> > $ g++ -xc++ -std=c++11 -dM -E - < /dev/null | sort | grep ANSI
> > #define __STRICT_ANSI__ 1
> >
> > $ g++ -xc++ -std=gnu++11 -dM -E - < /dev/null | sort | grep ANSI
> >
> > $
> >
> >
> > g++ with -std=gnu++11 does not define __STRICT_ANSI__ while g++ -std=c++11 does and those previous printfs functions and only those should be defined for c++11. The other traditional stdio functions should be defined only for g++11 and for traditional C.
> >
> > People that does not care about compliance will use the gnu++11 version while those that cares about compliance must use the c++11 version and so should be aligned the printfs series from stdio.h
> >
> >
> > Here at http://www.cplusplus.com/reference/clibrary/ is described which functions borrowed from C are included in the new c++11 standard. I worked for now only on the stdio.h functions but all includes should be checked against the strict compliance for c++11.
> > zos
> GNU++11 is C++11 with extensions.  I don't see a problem with having the
> i functions in that case.
There is no problem for having the i functions for gnu++11... but not 
for c++11

Just to emphase on the compliance for c++11: snprintf is part of the 
c++11 standard while fprintf, fscanf, printf, scanf, sprintf are not..

zos
> R.
>
> > > Corinna
> > >
> > >
> > > P.S.: For the time being, I removed the fcloseall function from this
> > >         declaration block and made its decalration dependent on
> > >         __GNU_VISIBLE.