[Mesa-dev] c99 vs v90

[Enrico Weigelt, metux IT consult]

On 04.08.2016 11:38, Jose Fonseca wrote:

> __FUNCTION__ has been around for a very long time (before 1999), so it's
> no surprise that MSVC has it.  But not implementing __func__ does seem a
> oversight.  I believe they fixed it on MSVC 2015.

hmm, okay. as long as we're supporting msvc'03, I'd suggest moving
everything to __FUNCTION__, which seems to be the most portable
one right now.

When, one day, we move up to msvc'15, we can reconsider moving to
__func__.

> To be clear, GNU C++ is the non-standard one here.  MSVC is following
> the spec.  It's just a ptty that the C and C++ standard bodies don't
> strive for consistency.

Now I'm getting confused ... C++ standard has __func__, but C doesnt ?

>> by the way: what's the exact reason for using msvc instead of gcc on
>> win32 ?
> 
> GCC on Win32 (MinGW) is not reliable (lots of known bugs.)  But above
> all, it's impossible to debug Mingw executables on Windows: one has to
> use GDB to debug Mingw DLLs/EXEs because Microsft debuggers don't
> understand DWARF debug info, and one has to use Microsof debugger to
> debug Microsoft DLLs/EXEs.  So it's impossible to debug an app that uses
> both, which is always the case if the OpenGL driver is built with GCC.

hmm, sure that this is still the case ?  (I knowledge on win32
development quite limited - last time I actively did that must be
decades ago ...)

> But Microsoft has been working on a Clang/LLVM frontend for their
> compiler, capable of genering debug info on MS format.  So I have hopes
> we could one day switch to it.  It still in its early infancy though.

hmm, what about building w/ clang anyways ?

>> can I also deploy it to some VM there and log in via VNC ?
> 
> No, not that I know of.

hmm, so I can't actually test the win32 builds :(


--mtx