[futurebasic] Re: Re: Re: [FB] "Genuine compiler"

Message: < previous - next > : Reply : Subscribe : Cleanse
Home   : September 2000 : Group Archive : Group : All Groups

From: BMichael@...
Date: Tue, 12 Sep 2000 20:50:37 EDT
>>   I'll send it to you privately (so I don't post files on the FB list).
>hey tedd, when you get that generic fractal routine,
>could you post it here to the f.b. list, please, i'd like to see it.

Geoff, if you'll send it to me, I'll post it on the futurebasic.org site; 
or, you can put it in the "dropbox" yourself:

<ftp://ftp.futurebasic.org/futurebasic/dropbox/> - then just email the 
list and let us know it's there.

On benchmarking... somehow, I think I got "volunteered" to come up with 
at least a minimal suite of relevant algorithms and mini-projects for a 
compiler comparison. Not just FB-RB, but more general. Anyone with ideas 
(preferably _short_ routines, real-world useful, and plenty of them - 
that's more meaningful than one huge monster, and more likely to get 
done) please email me off-list.

On going x-fb; I think as long as we're being this civil, and the 
interest is here for the project, we can continue to discuss it here. I 
_would_ prefer the "details" go on behind the scenes, either privately or 
on the x-fb list, just to keep traffic down to a low roar, and we can 
post the outcome here.

Geoff said:
>And again let me point out that I wasn't comparing REALbasic to FutureBasic
>nor was I really even comparing it to CodeWarrior. A guy on a newsgroup was
>stating that REALbasic is intrepreted. I was simply correcting him and since
>the message was about performance, I was pointing out that you can get great
>performance from RB. In fact, we very, very rarely get negative comments
>about the speed of REALbasic. That doesn't mean that REALbasic creates the
>fastest compiled code. What it means is that the execution speed seems to
>meet the needs of the user.

For some purposes, raw assembler isn't fast enough, at least on _my_ 
hardware. OTOH, for much of what I do, AppleScript or HyperCard (neither 
known for blinding speed) would be perfectly adequate. Speed is only one 
of many criteria to use in picking a tool. I use a language every day for 
a living that is known for it's ease of programming; can it compare to C, 
Basic, or even COBOL on speed? No way! Still, with the hourly rate of 
programmers, compared to the cost of a faster processor, and with the 
need for many, many hours worth of custom coding, a lot of companies are 
perfectly happy to fall on the "speed of coding" side of the equation. 
RB's database access, Windows code generation, and VB compatibility give 
it the edge in some areas; FB's speed, ease of toolbox access, and code 
size give it the edge in others. Whoopie! They're both useful tools; I 
just happen to be more "comfortable" coding in FB, as I'd imagine most of 
us here are.

Someone (or several) can take Geoff's fractal code and come up with an 
optimized FB version. (Geoff, can you send _compiled_ versions from CW & 
RB too? Otherwise we'll have hardware differences...) Meanwhile, we'll 
work on coming up with some other "challenges". I heard Scott Spencer 
volunteering to produce Object Basic versions, and I'll toss a glove at 
JavaSoft and Metrowerks... :-)

P.S. I actually saw water falling from the sky today! It was only over 
North Dallas, and only 1/10 of an inch, and lasted about 20 minutes - at 
first I figured somebody's sprinkler had gone haywire - no, it was a real 
"sprinkle"! Of course, there was zero "official" rain (they measure at 
the airport) so our record-breaking drought continues. The cloud cover 
did drop the temperature down under 100 for a little while. I don't care 
so much for myself - I've seen this "rain" stuff before. But it hasn't 
rained a drop this school year, and some toddlers may have never seen it 
at all...