Subject: Re: Binary releases and 64 bit off_ty
To: None <email@example.com>
From: None <firstname.lastname@example.org>
Date: 04/04/1994 17:21:41
> > The > 2G drives are rapidly coming down in price. I can see within a
> > year where it would be possible to need such a thing.
> Hey, even if so, wouldn't you rather make several partitions on such a drive?
> Anyway, I still think this HAS to be an option, only a (small) minority of
> our user base will ever face such decisions...
Large database applications would require > 2G filesystems...
> > Let's see what the actual result is before throwing it out.
> Sure, just make the extension conditional...
As Chris pointed out this would require ALOT of changes in areas
outside the amiga tree to support. It would be difficult to get
such changes in to the non-Amiga area of the tree.
> > I personally am against changing the default on something this
> > fundemental vs the other NetBSD ports. This is supposed to be
> I don't rate it THAT fundamental, it is one aspect of 4.4, nothing more.
Actually, changing the size off off_t does impact alot of areas
all over the OS tree. It's non-trivial to have #ifdef's in all
sorts of places, the main maintainers would probably disagree with
the attempt. If all the other ports can handle it why not us?
> My opinion here is: look at each feature, rate its usefulnes, what
> will it add to the system, what are the drawbacks. If the pros
> outweight the cons, go for it, if not, nuke it. I don't care VERY much
> about the name of the resulting system, I think it would be nice if we
> could continue the very close relationship to the other ports, but I
> (with my very own personal kernel :-)) will probably go my own way if
> NetBSD was to develop unconditionally into 4.4. Since we have source to
> all systems, this decision doesn't look as tough to make as if we'd be on
> a binary-only dependency.
I guess this is where the biggest difference is. I see it as
being more important that the Amiga port match the other netbsd ports
as closely as possible so we can use as much of the goodies
ported to NetBSD as possible. By picking and choosing features
we then have to create our own setup for various pd packages, i.e.
a touch of NetBSD, a smidgin' of SunOS m68k, a pinch of 4.3, a dab
of 4.4 and a sprinkling of ???. I'd rather do a
than have to tweek the gobs of net software for this feature left
out and this one changed slightly. You could always use the netbsd
config as a base but why bother changing it from netbsd? Seems a
waste of effort when there are better things to do then add
special case ifdefs, only for Amiga, over the kernel, library and bin
As Ty said, let's see how bad performance really gets. The 386 FPUless
guys aren't bitching about crappy performance so that might not be
an issue on smaller Amiga setups.
It would also be interesting to see if the kernel could handle both
sizes as is claimed by the core BSD team. I.e. program for 64 bit
but still be able to run old 32 bit code.