Re: Avatar

[gregt+@CMU.EDU (Gregory Ross Thompson)]

exuhag@exu.ericsson.se (James Hague) writes:
> This is part of what I meant when I said that the Avatar seems to be
> ignoring traditional 'wisdom' and starting over fresh.  I think we
> all realize that HLLs are preferable to assembly, at least when ease
> of programming is the concern.

  And portability.  C code is pretty much portable.  Assembly isn't in
any way, shape, or form.

> But if you look at what has happened
> to microcomputer software since HLL programming has become standard
> then it is not at all clear thar this is any kind of advance.  Now
> we have 12 meg word processors that are slower than their text mode
> equivalents.  Games have become less compelling as lazy programmers take
> lazy approaches to designing them.  And so on.

  Aaah, do you think someone's gonna code up WinWord 2.0 in assembly?
Think again.  It took a team of 20+ people a LONG time to write that.
It's taking even more people even more time to write 3.0.  If they
weren't using an HLL, it would never happen.  It's just that simple.

> Well, for starters, having all graphics operations offloaded to another
> processor is going to make a world of difference.  Second, a C compiler
> is not as important as you think.  If you want to be able to run tex or
> other packages, then definitely--and I'm sure some quality C compilers
> will become available.  But there are alternatives.

  No, there aren't.  If the machine is going to succeed at all, it'll
need a software base.  The only way it's gonna get a software base is
if people put the time into writing software for it.  If they have to
use assembly, 1) less people will want to program it, 2) it'll
take 'em longer to write app X, and 3) they'll have to re-write
EVERYTHING from scratch.  No current software company will ever
support the machine if they have to do all their work in assembly.  C
compilers are advancing rapidly.  There are some really nice
compilers out there.  The MIPS C compiler rocks.  The ORCA C compiler
sucks rocks (NOTE: I haven't gotten my update to 2.0 yet, so it might
not suck quite as many rocks, but it'll probably still suck in
comparison to other more advanced compilers.  NOTE2: Just because I
say it sucks doesn't mean I don't like it.  It's the only one we have
right now, so I use it, and I love it.  I mean no slam on ByteWorks.
In fact, I hope Mike keeps on developing it, and making it better.)

> Have you
> ever seen Action! for the Atari 8-bit machines?

No.  Has the rest of the world?  No.  Is it being used on any other
machines?  No.  Will it ever?  No.  Does it allow you to easily port
software from other platforms to the 8-bit Atari?  No.

> It's a C-like custom
> designed HLL for the 6502 which gets very close to the speed of raw
> assembly language.  Or what about a specially designed BASIC, like
> QuickBASIC or PowerBASIC on the PC?  They provide top notch programming
> environments IMO.

  Granted.  But hey, who uses 'em?  I'd like to have a good desktop
word processor, and a good desktop page layout program.  Are there any
on the PC or Mac written in those languages?  I don't think so, but I
may be mistaken.

> And finally, assembly language programming is a
> real breeze if you've got quality tools.  Assemblers on Unix systems
> and the PC are, for the most part, poorly designed by people with
> a severe distaste for assembly language. 

  Hmm.  I've done a bunch of assembly programming with decent tools,
and I didn't like to too much.  I did it 'cause I was getting paid to
do it.  I didn't think it was a breeze.

> IMO, the Avatar will not be the computer for everyone.  If you revel
> in the old computer industry saws ("faster is always better," "assembly
> language is too hard," "C is the ultimate programming language") then
> why not get a PC or a Sun?  The Avatar is a bold experiment, one doomed
> to fail.  But a bold experiment, nonetheless.

  I agree about the Avatar being bold & such.  I also agree that
today, C is pretty close to ultimate.  It all depends on the compiler.
If you write a good compiler (it's tough, but it's been done), then
you'll get good code.

  Well, I think I've been enough of a bastard in this post.  I think
that the Avatar has the potential to be a really cool machine.  I
don't have enough details on it yet to make any judgements on how good
it'll actually be.  I hope it does well enough to justify the effort
that Bill is putting into it.

		-Greg
		 InterNet: gregt+@cmu.edu
		       or: gt0t+@andrew.cmu.edu
		 BITNet  : R746GT0T@CMCCVB
		 AOL     : GRThompson