Subject: Re: Binary releases and 64 bit off_t
To: None <amiga-dev@sun-lamp.cs.berkeley.edu>
From: Ty Sarna <tsarna@endicor.com>
List: amiga-dev
Date: 04/05/1994 03:37:58
In article <9404050113.AA00552@icecube.cryogenic.com> billc@iceCuBE.cryogenic.com writes:
> Two issues,  #1 off_ty,  #2 FPU default.

That should be "off_t". I think "off_ty" happened when I hit y at the
subject instead of confirmation prompts. You might say it was a TYpo. :->
(double pun!)

> 1) My stance on this is based on the fact that what we have works, and that  
> adding something like this could change it.  Wait a sec, read on first, count  
> to 10, breathe deep.  I want this marked as a base, and that this base is  

Well, -current is, by nature, constantly changing, and it will continue
to do so. There are many major changes ahead before 1.0 happens, some of
which will doubtless cause more trouble than the off_t change. (eg, I'm
sure all hell is going to break loose when 4.4-Lite arrives...)

> When we're still getting unexplained panics, and reliability problems, we're  
> going to add in low level changes?  I'm all for Rambo hacking, but we need to  
> make sure that the newbie user isn't a casualty of this.

I'm not sure here wether you're pro- or anti- off_t changes here.  If
you want to preserve the status quo, at least for now, that imples you
want to *keep* the off_t change, at least for now.  The change has already
been made, attempting to undo it or make it optional will only add to
the instability. It's wonderful that the Amiga port is now being
maintained on sun-lamp, is current with -current, etc. I think it would
be a bad idea to diverge again and losse these advantages. It also means
more work when someone decides to sync up with -current again.

> 2) Thinking that most every CPU with an MMU has an FPU, is not reasonable.   

Note that I didn't say it was OK, just that it wasn't as bad an
assumption on the Amiga as for other plaforms (like the 386, where
almost nobody has 387's, and many 486's are 486sx's)

> part.  There's the GVP boards for the 2000, though I can't recall if they  
> offered the FPU as an option or not.  The 3000, and the 2500 machines  

The FPU was not an option on those boards, as far as I know... it was
standard equipment.

> (2620/2630) came with FPUs as default.  I can see no reason why someone with  
> the adequate CPU, memory, and harddrive space shouldn't be allowed to run  
> NetBSD on an A1000, just because they don't have an FPU.

I agree.  I also see no reason why someone with large drives shouldn't
be allowed to use them as well :-).  (Heck, percentagewise the no-FPU
minority is shrinking while the big-disk minority is growing!)

-- 
Ty Sarna                 "As you know, Joel, children have always looked
tsarna@endicor.com        up to cowboys as role models. And vice versa."

------------------------------------------------------------------------------