Re: Don't document the types of standard typedefs

["Carlos O'Donell" <carlos at systemhalted dot org>]

On Sun, Feb 26, 2012 at 7:31 PM, Joseph S. Myers
<joseph@codesourcery.com> wrote:
> Given the conclusion in
> <http://sourceware.org/ml/libc-alpha/2012-02/msg00454.html> that "GNU
> system" references should change to use various Texinfo macros
> depending on exactly what we think is being referred to in each case,
> I propose this patch which does some preliminary cleanup by removing
> "GNU system" references where I think removal, not converting to a new
> macro, is the right fix.
>
> The text removed describes the types to which various standard
> typedefs are supposed to be equivalent on the GNU system. ?To the
> extent that this is accurate for GNU Hurd it's mainly because of the
> lack of a 64-bit port. ?Since users should avoid relying on the
> equivalence of these typedefs to any particular type, I think it's
> most appropriate not to document those equivalences. ?(For off_t, the
> reference to fpos_t is simply wrong for glibc; fpos_t is a struct,
> necessarily so to include an mbstate_t.)
>
> 2012-02-27 ?Joseph Myers ?<joseph@codesourcery.com>
>
> ? ? ? ?* manual/filesys.texi (mode_t): Do not mention what standard type
> ? ? ? ?this is equivalent to.
> ? ? ? ?(ino_t): Likewise.
> ? ? ? ?(ino64_t): Likewise.
> ? ? ? ?(dev_t): Likewise.
> ? ? ? ?(nlink_t): Likewise.
> ? ? ? ?(blkcnt_t): Likewise.
> ? ? ? ?(blkcnt64_t): Likewise.
> ? ? ? ?* manual/llio.texi (off_t): Do not refer to equivalence to fpos_t
> ? ? ? ?or long int.

Looks good to me, we shouldn't be talking about this at all in the manual.

I don't understand your meaning about a lack of a 64-bit port for GNU
Hurd being the reason behind these descriptions? Could you explain
that in more detail?

Why would we document the native type of the standard typedef? Doesn't
that lead user to use the native type instead of the portable typedef?

Cheers,
Carlos.