🚀 go-pugleaf

In article <mwalker-250193094209@mwalker.npd.provo.novell.com> mwalker@novell.com (Mel Walker) writes:
>IMHO, the same could be said about GNU software: It does not succeed on its
>technical merit, but because it is free and widely used (i.e. marketing and
>standardization). 

I don't know about anybody else out there, but I can list (after
looking them up in the archives) a large number of GNU programs that I
haven't bothered to install here because we have no need for them (in
most cases) or they aren't any good (in a very small number of other
cases).  In other words, GNU software that has little or no technical
merit (which is extremely rare) does *not* succeed.

>Therefore, the same argument applies: How much would one
>pay for a slightly better compiler than GCC? 

Ask me first if I *need* a slightly better compiler than GCC.  The
answer is no; GCC is more than good enough, and blows most of our
vendors' compilers out of the water.  So therefore, I wouldn't spend
anything at all for a compiler which was only a little better than
what I have now.

>How much would one pay for an
>unbelievably better super-compiler?

The same principle applies.  (Quite frankly, I seriously doubt the
possibility of such a compiler on the architectures which we use
here.)

>Enough to even justify development
>costs?

I don't know.  First you should find out who needs a super-duper
mega-optimizing C/C++/Objective-C compiler.  Remember market research?

>How much incentive does a company that makes "free" compilers have
>for innovation?

You might ask Intel, Motorola, AMD, Data General, NeXT, and OSF.  Even
better, actually ***READ*** the SOURCE, and see for yourself.

> Almost none. Note: porting is not innovation.

Wrong, on both counts.

>Yes, it probably will force entrenched obsolete software out the market.
>Normal market forces do that, too.

Can you say MS-DOS?  I knew you could...

>The problem with GNU is that it forces
>other companies out the market.

So?  If they can't make a compiler that's better than GCC, they
shouldn't even bother.  Ask the people who are actually trying (and in
some cases succeeding) at doing so, and see if they agree.

> This reduces the base of those with
>incentive to make the new software and spur innovation.

So long as there are people performing basic research in compiler
technology, there will be an incentive to actualize that research.  So
long as there are people and companies who need to make that one
function run just a few seconds faster, there will be people doing
that research.  Companies don't create new and innovative software,
real people doing real research do.

>Clearly, the only reasonable alternative under a GNU-utopia would be to
>take jobs in software maintenance.

In case you haven't noticed, that's what most programmers are doing
right now.

-GAWollman

-- 
Garrett A. Wollman   | Shashish is simple, it's discreet, it's brief. ... 
wollman@emba.uvm.edu | Shashish is the bonding of hearts in spite of distance.
uvm-gen!wollman      | It is a bond more powerful than absence.  We like people
UVM disagrees.       | who like Shashish.  - Claude McKenzie + Florent Vollant