From:	Andrew Crabtree <andrewc AT typhoon DOT rose DOT hp DOT com>
Message-Id:	<199801281714.AA003507645@typhoon.rose.hp.com>
Subject:	Re: iostream concern
To:	eliz AT is DOT elta DOT co DOT il (Eli Zaretskii)
Date:	Wed, 28 Jan 1998 9:14:05 PST
Cc:	djgpp-workers AT delorie DOT com
In-Reply-To:	<Pine.SUN.3.91.980128131634.3916J-100000@is>; from "Eli Zaretskii" at Jan 28, 98 1:17 pm
Reply-To:	andrewc AT rosemail DOT rose DOT hp DOT com

 
> > I'm not sure if the headers came with the actual libraries of with
> > the compiler.
> They come with the libg++ distribution.
Do they still though?  I don't have the gcc 2.8 sources so I'm basing 
my guesses off of egcs.  Egcs integrates almost everything into 
one package.  C/C++/ObjC/G77, libstdc++/libio, makeinfo, collect2,
testsuites.  I'm still looking to see which part of it 
generates the _G_config.h file.  As far as I can tell right now
the libio config script just runs sed over an already existing one,
so I was assuming it was installed elsewhere. 

I believe that the libio for gcc 2.8 is just taken directly from
the egcs sources, but I'm not 100% sure.

> If I understand correctly, the problem with stddef.h is another
> problem, not directly related to _G_fpos_t, right?
Correct.  

> If so, please describe it.  
According to Jeff Law, unless you explictly change the configure 
scripts not to do it, the install process will always use and install 
its own version of stddef.h.  

> AFAIK, the headers which GCC brings with
> it are only for platforms where you need to compile it with the native
> compiler which doesn't have ANSI headers such as stddef.h.  This is
> not our case, so it shouldn't be a problem.
Apparently, gcc wants to do this regardless, in case gcc changes dictate
a need for a new version of the header.  It also runs fixincludes
even header files fixed by an earlier gcc release.

- Raw text -