This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: RFC: Mirror system header exposure in libstdc++-v3's headers
- From: Loren James Rittle <rittle at latour dot rsch dot comm dot mot dot com>
- To: libstdc++ at gcc dot gnu dot org
- Cc: bkoz at redhat dot com
- Date: Thu, 10 Apr 2003 18:07:00 -0500 (CDT)
- Subject: Re: RFC: Mirror system header exposure in libstdc++-v3's headers
- Organization: Networks and Infrastructure Lab (IL02/2240), Motorola Labs
- References: <200304050358.h353wSls047402@latour.rsch.comm.mot.com>
In article <20030407110834 dot 1c8282cc dot bkoz at redhat dot com> bkoz at redhat dot com writes:
>> Index: include/c_compatibility/math.h
> Does this mean that you are going to be using the include/c includes as
> well? That would be very interesting indeed.
Well, not really, I inspected all candidate locations based on
find/grep output. I did not consider whether a region of headers was
actively used. Sorry to get hopes up.
>>+ #if !defined (_GLIBCPP_USE_C99_EXPOSURE_HOOK) ||
> _GLIBCPP_USE_C99_EXPOSURE_HOOK
> Can you consider _DYNAMIC or _VARIABLE instead of _EXPOSURE_HOOK?
OK, _DYNAMIC it is.
> Other than that, I don't have any objections to what you are doing. Do
> you think it will work the way that you are expecting (the lldiv_t bits
> are the only thing that give me pause.)
Testsuite results improved the way I expected. Regarding lldiv_t, the
issue is that the system headers just don't expose the type unless
macros are set "right".
> I'm assuming you would compile the library with this macro set to 1, and
> then allow people to switch it off if they are feeling pedantic. I would
> think this would work.
Yes, the _DYNAMIC macros are effectively set to 1 when the library is
built. I will document that as a requirement.
> Hmmmm. Interesting.
Well, as usual, active need is the mother of invention. ;-)
I will post/commit a final, tested form with Phil's and your input.
Thanks,
Loren