Subject: Re: Protecting the rest of the kerner headers against multiple inclusion
To: Simon Burge <simonb@netbsd.org>
From: Eduardo E. Horvath <eeh@one-o.com>
List: tech-kern
Date: 09/11/1999 14:15:56
On Sat, 11 Sep 1999, Simon Burge wrote:

> Christos Zoulas wrote:
> 
> > The proposal is to have the last component of the path for MI code
> > and the machine name for MD code.
> > 
> > MI: _SYS_TYPES_H_ _NETINET_IN_H_
> > 
> > MD: _I386_TYPES_H_ _I368_ENDIAN_H_ *OR*

I prefer this one.

> > MD: _MACHINE_TYPES_H_ _MACHINE_ENDIAN_H_ 
> > 
> > I think that the first is better for MD code, the disadvantage being that
> > you will be able to include MD files from multiple archs... But that is
> > screwed up anyway, and should be allowed if you know what you are doing.
> 
> Maybe _MACHINE_ARCH_TYPES_H and _MACHINE_TYPES_H of you go with the
> latter MD idea - things in both arch/mips/include and arch/pmax/include
> come to mind.  I think the former MD choice would look better...

There are times when you have one architecture that's a superset of
another and you would want to include the same file from multiple archs.
For instance if you have a 64-bit arch and a 32-bit arch and you want the
64-bit arch to be able handle binaries from the 32-bit arch, etc.


=========================================================================
Eduardo Horvath				eeh@one-o.com
	"I need to find a pithy new quote." -- me