Subject: Re: General feeling on merging APIs...
To: Warner Losh <>
From: Jeroen Ruigrok/Asmodai <>
List: tech-kern
Date: 06/27/1999 10:11:59
* Warner Losh ( [990627 09:02]:
> In message <> "Justin T. Gibbs" writes:
> : The hope is to provide a consistent interface across
> : all *BSDs which is why I've addressed this to all of the *BSD projects.
> More generally, I'd like to identify areas where the kernel APIs of
> the various projects have diverged and see what, if anything, can be
> done about them.  I'd also like to get commitment from the various
> controlling bodies of the *BSD projects to a statement something along 
> the lines of:

Nice to see interest about this, it means a lot to me that the three BSD's,
all with their own fun in use, etc, are willing to work together on this

Also, what I am doing with the PDP (here he goes again) is that I am trying
to determine which BSD got which function where and in what version.
Would make an excellent reference for people wanting to get something
running on all three BSD's. (btw, ignore any FreeBSD words in the PDP, I
am currently making it *BSD.)

> 	We support the identification of API differences between the
> 	various BSDs.  We generally support the idea of merging the
> 	APIs, but reserve the right to veto anything too radical.

Sounds like a nice stance, however (not saying you did Warner) it _is_
absolutely fundamental that we, as the three BSD's speaking, put back
ego's where it's technical excellence that counts. Private agenda's never
served someone. So I myself am counting on the maturity of the participants
to discuss technical internals and not moving to ego-talk.

Note, this isn't about merging all the three BSD's back together. It's
merely about making the API not too diverse/different from each other.
(together we stand, divided we fall - cliche, but true)

To clue in the OpenBSD people who haven't been reading tech-kern of NetBSD,
it started when I pointed out to Bill Sommerfeld that FreeBSD already had
an asleep function, which he was about to implement. A good function, but
substantially different from FreeBSD's asleep, that I didn't think it was
a good idea to just let Bill continue without warning him at least, as well
as Matthew Dillon (original writer of FreeBSD's asleep()).

Can we say at least, that the core 4.4BSD functions have remained largely/
almost the same across the 3 BSD's?

Can we point out at least a few areas in which we differ?

Are there differences spotted at hand which can be resolved on short time?

> I've been trying to put together some compatibility shims for
> newconfig drivers in FreeBSD's new-bus world and have run accross a
> number of minor differences that would be trivial to fix.

That sounds like a good thing Warner. Any more to tell about this wrt NetBSD
and OpenBSD?

Looking forward to hear from the lot of you,

Jeroen Ruigrok van der Werven                asmodai(at)
        The *BSD Programmer's Documentation Project 
Network/Security Specialist      <>
*BSD: We are back and will not accept no...