[Uk-lispers] Meeting Wrapup
hibou at onetel.com
Fri Dec 9 17:39:43 PST 2005
On Friday 09 December 2005 12:35, Noel Welsh wrote:
> My thanks to Shriram for not one, but two excellent tasks,
> and to everyone who attended.
> I, unfortunately, had to
> leave early but I trust the conversation carried on in the
> pub and a good time was had by all.
> There was a discussion about language popularity, and 1)
> I've got my own opinions on this topic, and 2) I didn't get
> to bore you with them at the meeting, so here they are:
Here's my three-ha'pence worth. There are two issues here
(1) I think there's a trade-off between popularity and quality,
and this is unavoidable. It's not just with programming languages,
it's with things like food, beer, movies, music, art. Improve the quality
and you lose customers (the Citizen Kane end of the spectrum).
Of course, lower the quality below a certain threshold and you also
lose customers (the Police Academy 6 end of the spectrum). Aim
at the centre of the bell curve (Star Wars, etc.) and you can make big
So if you want to make Lisp (in which I include Scheme) popular, you
have to make it worse. This has, arguably, been done already and
Ruby aims somewhere between Java and Lisp, and you can expect
it to be less popular than Java, and more popular than Lisp. If it
improves, it will lose market share.
The only hope is that if you educate the population sufficiently, their
taste will improve. I think this has happened with food. Macdonalds
(which aims at or slightly below the centre of the bell curve) is
improving the quality of its food (albeit from quite a low base), because
staying the same would cause it to lose market share. (This may be a
purely UK-centric observation.) On second thoughts, this is more
likely to result in improvements to Java and VB, rather than programmers
abandoning them for better languages.
There is still one commercial advantage in being good: good stuff tends
to last. People are still watching Citizen Kane, listening to JS Bach
and reading Homer. Similarly, they're still using Lisp. (It's a very
different language from LISP, but the core features -- program/data
equivalence, symbols, linked lists, functions as first class objects --
are still there 45 years later.)
(2) There is another issue, which you recognize: if the language is very
good, but impractical, even programmers with taste will steer clear
of it for practical tasks, only to wheel it out for the occasional
demonstration of its coolness.
So I'd expect there to be a standard distribution of any language/dialect
which supports the sort of libraries people need -- graphics (including
3d), sound, maths, Internet, relational databases (if you want it used
commercially), etc. They should be packaged up in the distribution
and work straight out of the box.
I'd also include libraries which are unavailable in VB/Java/Ruby/whatever,
ideally tasks which require symbolic programming, so that people who
need these features, and the generic stuff, have nowhere else to turn.
More information about the uk-lispers