[fpc-pascal]Kylix and M68k Port

Kuba Ober winnie at hoth.amu.edu.pl
Thu Nov 30 15:13:49 CET 2000


> The original post was quite lucid and well written, but I think you missed
> the intent of it. He was merely pointing out what most of us can plainly
> see; that we're a minority. Like it or not that fact is irrefutable. True,
> some people seem to follow the masses and use whatever is in vogue, but the
> numbers are certainly not in our favor. That doesn't make it right or
> wrong, it's just the way it is.
>
> I happen to love Pascal, and have been programming in it for longer then
> I'd care to admit, but that doesn't mean everyone else knows how great it
> is.
>
> > Go to the Borland Website. Search for the VCL scanner,
> > and run it on your machine. Then be surprised ;) Microsoft
>
> I tried this and came up with 6 programs and a handful of support files. I
> have well over 500 executables that run something, be it GUI or command
> line. That's not good odds...
>

But alas, whether we're minority or not, I used to program everything in 
Pascal. Started with Turbo Pascal 3 on CP/M which was a breaktrough those 
times, especially that I was an elementary school kid back then and that 
small bit of `visuality' that they offered was very, very nice. Then I used 
it on PC, and iterated all the way up to BP7.

Then came Delphi, I tried it and I switched to ADA ;-). Looking from a 
greater perspective, Delphi can force you to code nicely if you're not a very 
loose(r) programmer, especially that the framework is very easy to use. But 
when it comes to stability of the code itself, and the stability of resulting 
executable, Pascal-alikes are decidedly on the top. There are several C-ish 
hacks that work in C++ as well (unless you devote your time to read those 
manpages and turn all the warnings on ;-), and there are simply too many 
people who use those, and this results in code that will compile on i386 but 
not with i686 optimizations due to code generator bugs, that will compile on 
sparc but not on rockwell because one is more `compliant' than the other, and 
so it goes and goes. Just by looking at the discussion lists and seeing how 
many build problems vs. useful issues there are raised, C/C++ lists are 
definitive leaders, and the bias is completely different in Pascal/FPC lists.

The aerospace industry has parts of the code in C/C++ (Hubble ground ops 
`backend' code as an example), even though ADA or Pascal would be much safer 
and easier to maintain. Yet it's only because there's abundance of C/C++ 
programmers out there. I never had any big problems with Pascal code 
portability. Most of my general-purpose code that wasn't I/O related ports 
easily across all Pascals, and I never ever had typical C/C++ hassles of 
cryptic error messages, you-dig-it changes to makefiles in order just to get 
it to compile, etc. People might not see it, but C/C++'es (both source syntax 
and compilers themselves) have many very-important-options (read archaisms) 
that were pretty useful on PDP machines when many users were trying to share 
what is nowadays present in mid-level pocket calculators, but even CP/M with 
64k's of RAM was large enough to allow you to easily write large things in 
Pascal. I don't know if anybody did any survey about this, but if one would 
compare the amount of time spent really coding vs. amount of time spent 
maintaining makefiles and build environment (really a project-unrelated 
stuff!), Pascal and ADA are definitive leaders and anybody who argues either 
is blind or is an expert, disciplined script coder and very fast manpage 
reader (in latter case hats off).

If one tries to discuss whether Delphi is better rad than any C/C++ thingo, 
there are tons of pro-Delphi arguments. I bet that Delphi (and Pascal and 
ADA)-using companies really benefit from that by lower costs and by staying 
on-schedule easier. Sometimes I think that writing a C++ front-end (name it 
FCC then ;-) to FPC with more strict syntax checking that would disallow 
C-hacks and integrate C++ into the unit model of Pascal (instead of ugly 
#includes and .h files) would cost some companies less time than trying to 
make their makefiles portable ;-).

Have tons of fun, and although I'm only occasionally downloading FPC to look 
at the progress, the real truth is out there and you're at it ;-))))

Kuba




More information about the fpc-pascal mailing list