This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: -std's (was Re: v3 link failures analyzed)
Gabriel Dos Reis wrote:
> I don't know if there is any evil shell which treats '//' as a comment :-).
$ whence c++sh? :-)
> | .. This thread originally started off because SCO
> | OpenServer presents a different set of headers in ANSI mode, so the headers
> | seen during configuration aren't the same headers seen during building,
> | or during the testsuite runs.
>
> Yes, I'm keeping that in mind. One intesresting thing to know is
> whether the system supplies different runtime libraries (I don't
> think so, but now I'm refraining from making any "reasonable"
> assumptions :-).
We should think about this in the abstract. Can you think of reasons
someone might provide different libraries for strict conformance vs.
common practice? It might provide different libraries with reduced
namespace. I don't know about -ansi vs. ansi-not, but there are
other similar flags (i.e. xpg4 vs. posix) that have different runtime
characteristics for the same services (sigh) and I could certainly
imagine systems slotting in different libraries behind our backs for
that.
I can't tell that my systems actually -do- replace the libs, though.
It assumes if you're specifying spec conformance (presumably becuase
you're a masochist) that you'll pay attention to those missing prototype
warnings...The runtime thing is handled by setting a flag in startup
code and testing it in the libraries.