| I believe the current WG14 charter is here: https://www.open-std.org/jtc1/sc22/wg14/www/docs/n2611.htm This text is still in force, it seems: “
13. Unlike for C99, the consensus at the London meeting was that there should be no invention, without exception. Only those features that have a history and are in common use by a commercial implementation should be considered. Also there must be care to standardize these features in a way that would make the Standard and the commercial implementation compatible.
” I read this as saying that anything that gets standardized should be available in one of the major implementations. In practice, most of the qualifying features will have been implemented in both GCC and Clang in the same way, so for most users, there is not much benefit from standardization. Some may feel compelled to support the ”standard” way and the “GCC/Clang” way in the same sources, using a macro, but that isn't much of a win in most cases. Of course, there will be shops that say, “we can't use feature until it's in the standard”, but that never really made sense to me. Things are considerably murky on the library side. In my experience, library features rarely get standardized in the same way they are already deployed: names change, types change, behavioral requirements are subtly different. (Maybe this is my bias from the library side because I see more such issues.) For programmers, the problem of course is that typical applications do not get exposed to different compiler versions at run time, but it's common for this to happen with the system libraries. This means that the renaming churn that appears to be inherent to standardization causes real problems. Others have said that new standards are an opportunity to clarify old and ambiguous wording, but in many cases the ambiguity hides unresolved conflict (read: different behavior in existing implementations) in the standardization committee. It's really hard to revise the wording without making things worse, see realloc. So I'm also not sure what value standardization brings to users of GCC and Clang. Maybe it's different for those who use other compilers. But if standardization is the only way these other vendors implement widely used GCC and Clang extensions (or interfaces common to the major non-embedded C libraries), then the development & support mode for these other implementations does not seem quite right. |