Shriram Krishnamurthi <shriram@new-world.cs.rice.edu> writes:
| Perhaps this is true of Lisp. There certainly are middling-to-small
| Scheme implementations. SIOD has looked tiny (to the casual glance);
| Rob Warnock has sung its praises many times, especially as a quick
| start-up low-weight platform.
I like SIOD a lot, but it isn't typical. The next larger group (elk,
guile, etc) are much bulkier.
| You seem to be arguing a structural notion (whether or not there are
| such things as distinct "scripting languages") by citing specific
| implementations that didn't see or chose not to occupy a niche. I
| don't agree with this line of reasoning.
|
| Surely I've misunderstood you?
I think it is true that some languages are more suitable for
scripting, on human factors grounds, based on syntactic simplicity.
That's not to say that any random language couldn't be used as a
scripting language, just that when less typing is required people will
be happier given an application domain where most programs are pretty
small anyway.
But that's not the point I was making before. Rather, I was observing
that Tcl was conceived with a goal in mind, and was designed and
implemented in such a way that the goal was achievable and efficient,
and that Lisp and Scheme implementors typically have different goals
and different implementation strategies, and the result is that
available Lisp and Scheme implementations often have a hard time
filling the same niche that Tcl does. An implementation could, in
principle, do as well or better, but somehow, in practice, they
usually don't. That's not an argument about lisp in the abstract, but
a complaint about the typical poor quality of implementation that I've
seen. And, as we said above, siod is a notable exception. Scsh
(since this thread got crossposted there) is a fine example of the
worst possible strategy, where an interesting design is spoiled by an
inefficient implementation.
Maybe that's just shallow and unhelpful, and if so I apologise for
wasting your time, but when tcl has been the basis of a whole bunch of
really cool stuff, and a number of people post articles to the effect
that Ousterhout's ideas are all wrong, it makes me want to defend his
design and his quality of implementation, because I think those sort
of practicalities make a difference. Before putting tcl down, take a
hard look at how much more successful than scheme it is, and notice
that this isn't by accident.
|