The USER_H issue

[Marc Espie <espie at quatramaran dot ens dot fr>]

In article <564.937826241@upchuck.cygnus.com> you write:
>  In message <199909151623.JAA02391@zack.bitmover.com>you write:
>  > > There are no GCC headers in my base system.  We've never needed them, and
>  > > there are copyright issues anyway.

>  > Do you have your own stdarg.h too?
>They may have their own stdarg.h, but we should not be using it.  We should
>be using the gcc supplied stdarg.h & varargs.h.  This is related to the who
>thread about how a port should never, ever override USER_H.

I'm sorry, but you keep saying overriding USER_H is wrong, even for
distributors, and I keep telling you that we will end up with two distinct
sets of headers, one for the system proper, and another `special' one for
gcc.

As far as the OpenBSD project is concerned, this is more efficient for us
to rely on our own headers, and tweak the system installation so that those
get used and debugged, instead of having two separate compilers with distinct
sets of bugs...

Please do tell me what's in it for us to remove the USER_H overrides.
I understand that this means simpler maintenance for the gcc project, but
it looks like it only has drawbacks for the OpenBSD project as far as
maintenance is concerned: we just get those `generic' headers with hundreds
of tweaks that don't apply to us, we will probably get some hard to debug
interaction bugs at some point.  It really looks simpler and more robust
to just rely on our own headers and make sure those work.

Not to say you can convince me, but you will have to use real, 
logical arguments, something more convincing than 
`overriding USER_H is wrong'.