[clean-list] 2. Re: Re: Communicating with other programs

[Siegfried Gonzi]

clean-list-admin@cs.kun.nl wrote:

> Message: 2
> From: "Marco Kesseler" <m.kesseler@aia-itp.com>
>
> > How important are the I/O stuff for the users?
> [...]
> > Lets be honest: inventing the wheel is stupid.
>
> Exactly.

I rather meant my idea of implementing a 1000+n plotting library is stupid. The idea behind a functional-style GUI
class is not bad at all, but nobody jumps onto the train.



> Yup. And one of the things I have been nagging about is simply to get some
> priority to (language) support for interfacing with others. Opening up the
> sources may be one way of getting this in case the clean team has not enough
> resources.

Currently, I am implementing a 2 dimensional Kriging interpolation (it will be ready in a few days, if you are
interested in, you can contact me off the list; you can also get a --undocumented--library for reading files as
follows:

header
Date,Time,Val1,Val2,...
01:04:2001,03:12:23,2.3445,23.4455
01:06:2001,04:12:45,345.3454,23.22
...

the values became stored into an array (and a side effect is that the day and time will be converted to the fractional
part of the year).

But I am tired in Clean of coping with unique arrays. I am not any longer willing  to strain my nerves with Clean's
unique array concept. It really works very strong against the programmer.

My next project is maybe in Ada 95, because there I can define a type which maps a number to a defined precision
(actually, in Clean I had to implement it myself for my 2 dimensional Kriging interpolation on grids).

And it is also unfair (with respect to other scientist) starting from my side to program in a language which nobody is
using. Surely, one has to make the first step, but this is only true if one can show (me) that the step really pays
off.

 I am tired of justifying my position to show my colleagues that Clean is better. But the tragedy actually is: Clean
is not better. I am not faster in developing my programs (even when compared to C), I have to implement too many "low
level" routines (thanks to Fabien T.; without his library for writing formatted output, Clean would be useless for
me).

Every day, I get a better understanding why the progress of Clean in the scientific community is depressing. It is not
necessarily the fault of the language, but I once red a comment nearby at an Ada pushing site. It has been about that
often in academia there is no understanding what people really need and it is often very true that also in academia
seldom anybody has ever took part in a big software project.

That said. There are many interesting aspects in functional languages and in Clean but maybe my programming tasks do
not need many of them (lazy evaluation, currying, passing functions as arguments). One thing that I will miss in Ada
(sometimes I need a control over numbers; yes I can implement in Clean a function which rounds ceil but I do not see
any reason why this has not became standard with the distribution itself; surely I can complain about that but I bet
the situation will not change with the Clean 2.0 compiler)  is pattern matching...


Regards,
Siegfried Gonzi