This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: std::isinfinite
Paolo Carlini wrote:
> Mateusz Loskot wrote:
>
>> May be I don't understand your suggestion, but if I do I see a few
>> issues: - clients of my software usually use system-wide
>> installation of compiler and libstdc++ - they expect to unpack
>> package and run commands triplet: ./configure, make, make install -
>> they are not interested in or not able to do manual customization
>> of dev environment, rebuilding libstdc++, etc.
>>
>>
> Now I don't understand: those names are in namespace std, which
> problems are causing to your software?
I'm sorry if I've confused you.
My objectives are:
1. How can I get it reported if a developer from my team will use
GNU GCC extension in our project?
2. How can I prevent from using GNU GCC extensions from "client side",
*without* rebuilding libstdc++ library?
3. The main question is how to achieve *strict* compilation mode,
reporting *all* non-C++ Standard (and C98) stuff, if any used
in source code of my project.
I hope it's clear now.
> Maybe we should fix that, if possible (usual difficulties with "C"
> facilities).
According to my sense of standard compliance, it's very incorrect to put
some non-standard stuff into std:: namespace.
It's like declaring user's type in std::, what practice is not recommended.
> In any case, consider that all those facilities are
> already in std for the next C++ standard, are rather uncontrversial
> C99 facilities, and it seems, well, late to implement complicated
> solutions...
Hmm, it's interesting. Could you point me any document about that?
I couldn't find anything explicitly said about it, or you mean
it's a subject of "Synchronizing the C++ preprocessor with C99"
http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2006/n1969.html
Cheers
--
Mateusz Loskot
http://mateusz.loskot.net