Subject: Re: Porting code
To: Brian Buhrow <buhrow@lothlorien.nfbcal.org>
From: Don Yuniskis <auryn@gci-net.com>
List: port-sparc
Date: 11/28/2001 01:54:32
> Don.  It's worth noting that if you're careful with your
>port to FreeBSD, the code "should just work" under NetBSD.  

Heh heh heh... "famous last words"?  :>

I think the big difference is that the FBSD camp has really
just focused on *one* port (i386) and "the rest be damned"
(though there is now an attempt to backfill to other
platforms... yeah, right.).  As a result, often hooks
exist and implementations rely on x86/PC specific things.

The NBSD camp, OTOH, has seemed to focus from the start on
making implementation decisions that tend to be more
"generic" (poor choice of word) and more readily extensible
to other platforms.  I don't have the experience to be able to
know the subtleties between them at various levels...

>Also, you could try bringing up
>NetBSD on an I386 box to do your port as well.  

Yes, I have one such box.  But it also already has most of
the tools I need installed under FreeBSD ("multiboot").
It's a PITA to go through and set most of those things
up and configure them for my needs on another system
(especially when they are already up and running)

>The programming
>environments under FreeBSD and NetBSD are really very similar, and if you
>keep the idea of running the code under both in mind when you do the port,
>I think you'll find you can produce a very portable port.

The bottom line is I need to do the port in order to finish
a "real" job/product.  The ideal is to just do the port as I
need it done, and then wrap it into a package (after dealing
with the license issues, etc.).  Having to make a second pass 
over it is just a nuisance (i.e. let someone else look at
my port and do that...).

I suspect, however, that rather than fighting with lawyers
over language terms, I will just let the vendor deal with
integrating my changes into their codebase and redistributing
them in their next release...

<shrug>  I have another project to port after that so
maybe I will have more time to "risk" on that one.  And, I
can spend the interim time tinkering with NBSD and seeing
if other hardware targets might be better to work on
before that time comes.  Talking to a friend about an
Ultra that might be freed up soon... <shrug>  "Toys"  :>

Time to go see how the citrus trees are faring (hard
frost expected tonight.  Grrrr...)

--don