Subject: Re: long long vs. -ansi...
To: Peter Seebach <seebs@solon.com>
From: J.T. Conklin <jtc@cisco.com>
List: current-users
Date: 04/08/1997 09:05:14
In article <199704080414.XAA16796@solutions.solon.com> seebs@solon.com (Peter Seebach) writes:
The solution is to hack gcc to provide a *pure extension* called, perhaps,
__int64_t, and use that for the quad type; then no code needs to know about
it, and we aren't producing syntax errors.
__int64_t invokes undefined behavior, but it's not a syntax error, so it's
fine to use it even in -ansi -pedantic.
As I mentioned in another message, I believe gcc already has an option
to allow non-ansi stuff in system header files. I thought we were using
it.
We may want to look at migrating to C9X-style <inttypes.h> support,
especially because the long long rules are changing as we speak, to
avoid some of the cruft inherent int most current implementations.
I believe that it would be a good thing to be an early-adopter of that
header. Unfortunately I dont have a C9X draft. Peter, would you be
willing to code up a inttypes.h we could integrate?
--jtc