True. I tried to muck with Tk to get it to work with another
language (gnu smalltalk), but it has too many Tcl assumptions.
- String only syntax. Everything's a string. Semantics change
depending upon where you are. Ie, comments can cause syntax
errors. What's worse, this comment problem is acknowledged,
but treated as only a side effect of the language rules,
NOT as a bug that will be fixed in later versions. Hard to
imagine a computer scientist wrote this (at least in Perl,
Larry Wall doesn't claim to be a computer scientist and only
wants a useful tool).
- Writing some small programs, or fragments callable from main
routines is easy. Writing larger programs with more advanced
data structures can get to be a nightmare.
- Hubris. Ousterhout treats this as if no one had ever thought
up embedded scripting languages before, and we should all thank
him for coming up with this lesser example of one. Ok, there
weren't many on UNIX, but that's not the entire world. Personally,
I thought REXX was nice, and the AREXX version made embedding it
TRIVIAL (it runs via a communication port, unlike Tcl where you
have to link all of it into your application). But no mention of
these when he gives talks, just the implicit notion that he was
the first. (Tk talks get just as religious) Of course, you have
to forgive him, coming from Berkeley where turning industry ideas
into academic papers is not unheard of.
-- Darin Johnson email@example.com "Particle Man, Particle Man, doing the things a particle can"