Comment by bsder
13 days ago
Because Lispers sneered at the microprocessors of the time while everybody else piggybacked on them. And, by the time that attitude changed in the early 90s, there were already socially entrenched languages that had the "superpower" of garbage collection (Tcl for ECAD, Perl for sysadmin).
Nobody in the Lisp world ever took the time to implement stuff that people wanted on those tiny machines. Or to demonstrate to people the cool stuff it could do.
You can see this in Dr. Dobbs Journal. People are doing things like drawing graphics, writing spell checkers and controlling modems. Assembly and BASIC are normal but C and Forth are mentioned regularly. Turbo Pascal pops up in 1984. Some of the names are famous enough that you recognize them even now, decades later.
Lisp just ... gets barely mentioned in passing sometimes. And nobody of note writes anything about it. Somebody could have built a word processor, a spell checker, a chess game, a reversi game, ANYTHING ... but nobody did.
For the record, Tcl doesn't use garbage collection, rather objects are reference-counted and freed when ref count is <= 0.
But interestingly, Tcl is a good example of languages that took a piece of Lisp's domain. In fact, Tcl is quite Lisp-like with a syntax distinct from Lisp. Tcl has been successful and with the recent release of v. 9.0, it stands to gain traction among programmers.
OTOH CL and Scheme remain underused though current implementations are generally well-equipped to handle contemporary requirements. I've used Scheme to build website generators and other tools but there's a dearth of large-scale, visible Lisp/Scheme projects out there to attract developers.
Starting such a project is a big commitment, probably programmers are hoping somebody else will pick up the ball and run with it.
Ref counting is garbage collection (cf https://courses.cs.washington.edu/courses/cse590p/05au/p50-b...). It's just not tracing garbage collection.
I'm a decent fan of both Tcl and CL, but Tcl has the big problem of being "almost" homoiconic and lacking good meta-programming tools like quasi-quoting. I say almost because comments break homoiconicity, whereas in CL they are discarded at read-time, never appearing in the parsed tree.
Quasiquoting is only necessary in Lisp because Lisp evaluates its arguments by default. Tcl does not do that.
1 reply →