> >> any?  If so, then there's a big world of computing you've managed to
> >> miss somehow.
> >
> > Hardware is cheap, software isn't.  A PC is a few of hours of your
> > hourly rate.  How much auto-router code does that buy?
> 
> Let's see... Imagine an auto-router that's used by 10'000 users. Let's
> say "a few hours" is 10 hours. That's 10 hours per user, so it's 100'000
> hours. How much auto-router code does that buy again? :)


Not much.  So you get 100 man strong team together and they code away for 6
months.  In machine code because that's the 'in' thing.  More speed!
Ludicrous speed even!  They develop a whole new algorithm that's 10% faster!
Meanwhile I'd upgraded my 4 year old PC (like Olin has) and have had a
better speed increase than 10% for that time.  So, new PC with instant
results now, or wait 6 months for something that's 100% machine code, 10%
faster and buggy.  Let me think it over...


> > There's a limit to the gains you can get with code, but you can buy
> > cheap CPU cycles right now.
> 
> Isn't there also a limit how many cheap CPU cycles you can buy right
> now? Where do you go when you have bought all the cheap CPU cycles you
> can buy and still are not there where you want to be?
> 
> Of course, from the viewpoint of a single user this may be the solution
> -- if no other solution is available. But between two autorouters with
> similar functionality, one may be twice as fast, and only a bit more
> expensive. So if such an option is available, which one would you pick?
> The slower product?


If the slower product has less bugs and better support, then yeah.  But
that's irrelevant, of course you pick the faster one if everything thing
else is equal.  And buy a new PC to run it on.


> > Or are you waiting for the C programmer to save the day with v4.56b?
> 
> You're good in setting up strawmen, or at least are trying to be. But it
> may be that you don't have to wait... there's competition out there,
> even for autorouters.


But the customers are locked in to their existing product.  No-one will
switch as you need to do all your libraries again.  But that's irrelevant
too.


> >>> C is rarely needed these days.
> >>
> >> That's better, although I think you meant to say the extra
> >> performance of a compiled language is rarely needed these days.  I
> >> agree with that, but note that this doesn't make a compile language
> >> a bad idea, only that it makes
> >
> > No, I meant C.  C & a HHL language can both compile to something
> > that's pretty much indistinguishable.
> 
> Indistinguishable in what sense?


To the user.

 
> > Microsoft showed that with QuickC & QuickBasic years ago.
> 
> What did they show? I tried to find something about this, but didn't
> find any relevant findings :)


I used to write in both.  The compiled QB programs ran just as fast as
compiled C ones, and TurboC & TurboPascal probably did too.  (Hence MS
dropping the compiler in VB to make C++ the 'better' choice.)  Sure it was
hard to write image manipulation stuff in QB as it lacks the bit-twiddling
stuff C has, but hardly anyone did that, despite many claiming to.  Write
that little bit in C and the rest in something easier to use.  

Or just buy a library where some other poor sod has done the hard work for
you.  Like Olin said, you don't write it yourself, but you still need to
know how to make the right choice.  Sorting routines are a good example.

Reminds that Olin mentions he had to write a program that took a long time
to do a lot of complicated stuff, but never actually mentioned what he wrote
it in.  Pascal?  ADA?  GWBasic?


> > Why write in a language that creates more coding errors?
> 
> Like for example assembly? Because in some cases and instances the last
> bit of control or optimization that a compiler can't give or do may be
> important.


More than one company went broke waiting for the assembler programmers to
finish while the HLL (even C counts there) guys beat them to the market.
Fast doesn't matter much anymore unless it's fast to market.  Some people
want their stuff to actually work too, which is rather novel.

The car analog for this thread is arguing whether your car needs a rear wing
on it.

Tony

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist