[Freeciv-Dev] Re: curiosity

[Gregor Zeitlinger <zeitling@xxxxxxxxxxxxxxxxxxxxxxx>]

On Sat, 1 Dec 2001, Alan Schmitt wrote:
> How about a fast, modern language, like Objective Caml.
It's interesting you brought up the point. Until Wednesday I thought Java
was state of the art in programming technique, but then I read a slashdot
article about lightweight languages. What caught my attention was that
the research academia basically said hackers are fools because they don't
know how to program in a proper manner. I then followed some links and
found that there are really interesing languages and some features were
just very appealing. Objective Caml was one of them, and one that was
ranked 2nd on a speed test after gcc and faster than g++. I thought that
this was really impressive. Even more impressing was that it works
without any variable assignment. All loops are done recursively without
any loss in speed (I guess the compiler makes it iterative). And anyone
who's programmed a declarative or functional/logic language knows that
some problems can be handled very neet that way.
I also found Ada which I liked for the very strong typing and
smalltalk and eiffel for their pure OO character. the latter one has a
very clever concept to test programs using pre- and postcoditions.

Having done that, I wondered if all people who use those deprecated
languages are fools, including myself.
I came to the conclusion that this answer was too easy and tried to do an
economical analysis of programming languages.
Since there is only a finite amout of time people are willing to learn
languages (and get good at them) there is a tendency towards fewer
languages (for the same purpose - and I don't mean that they are turing
comlete). Thus if a new hackers is born, he'll decide what language to
learn based on the likelyhood to use the language. That will be one of the
popular ones. He'll get used to it, and feel comfortable with it
(regardles whether it's good) and use that one or languages with a similar
paradigm. Thus the popularity of a certain paradigm can remain for a very
long time, even if it's not state of the art.
Another network effect is the availability of extentions/modules/packages
for a given language. That is, IMO, one of the major reasons, C++ and Java
are state of the art - not from a purely technical point of view - but
from an effective one, taking into account hacker and extention
avaliability.

If I had to choose now what language and everybody would follow me and
port all toolkits and extentions to it and would learn and like that
language, I would probably not choose C++ or Java - but you know how
likely that is.

-- 
Gregor Zeitlinger      
gregor@xxxxxxxxxxxxx