4 ms·

It's a little sad that it's declined so far. In a way, it's a victim of its own success--it's so easy to write a lisp system that literally dozens, if not hund

It's a little sad that it's declined so far. In a way, it's a victim of its own success--it's so easy to write a lisp system that literally dozens, if not hundreds, of variants sprang up. The community was divided, and it never quite came through. IIRC, our fearless leader PG made his zillions using Lisp.

jacobobryant 7y ago
> it never quite came through. Hard disagree. The Clojure ecosystem is great.
- downerending 7y ago
  It's obscure enough that it perhaps proves my point. Nonetheless, I'll have to look at it.
  - bcrosby95 7y ago
    If you haven't heard of Clojure, you probably haven't listened to a Rich Hickey talk. They're worth it even if you don't use Clojure. "The Value of Values" and "Simple Made Easy" in particular.
kazinator 7y ago
The division was never a problem during the heyday. The decline in Lisp in fact started at the height of standardization. > it's so easy to write a lisp system Have you tried it, and how far did you get? Making a Lisp system useful for production is pretty brutal. 13 years into it you will be debugging some eleven-year-old GC or compiler bug. People think it's easy because they visualize the job as being done when some executable file converts (+ 1 1) into 2.
- downerending 7y ago
  Maybe it declined for other reasons, but my impression was that the Lisp world was very splintered. I certainly can't think of any other example of a set of quite similar languages that's anywhere near as large. The Xerox Dandelion and friends was a thing to behold. If you ever got to touch one, you never forgot it. In my school, implementing a basic Lisp interpreter was part of a required class in the CS curriculum. (If it's not still, everywhere, well, then for shame.) Yeah, production is something else entirely. By "easy", I mean that dozens if not hundreds of people were tempted to write their own slightly better Lisp. Many of those did reach a production-ish level. But then what? There was no flag to rally around--just lots and lots of niche systems. I'm not sure. I think if the community had rallied around Common Lisp or Scheme, or maybe just those two, it might have ruled the world. It just didn't happen. And I'm miserably sad about that. What do we have now? Elisp is great, but I can't build production on emacs. I'll look at Clojure, but I'm dubious. Python is kind of lispy, but has had its own destructive schism. And C++ marches on--somewhere buried in there is a mildly functional lisp, using the worst of all possible syntaxes.
  - TeMPOraL 7y ago
    I'd think a refreshed Common Lisp would still have a chance. It's a great language with some really solid open-source implementations, but it got standardized at the transition point in our industry, so the spec doesn't even consider (now-commonplace) things like threading or networking, while exposing you to some abstractions over systems died out. The problems of CL wouldn't be unsurmountable if the community was larger, though. All the important stuff that wasn't standardized gets added by each implementation anyway, and then portability libraries get created that ensure consistent interface. But the state of most libraries is... rough at the edges. I contrast that with Clojure, with which I spent some months over the last two years. The library ecosystem (not Java-side, but Clojure-specific) is great, and a lot of care goes into it. I particularly remember being in awe of just how thorough Liberator is[0]. -- [0] - http://clojure-liberator.github.io/liberator/tutorial/decision-graph.html http://clojure-liberator.github.io/liberator/tutorial/decisi...
  - lispm 7y ago
    > I certainly can't think of any other example of a set of quite similar languages that's anywhere near as large. Scripting languages on Unix.

jacobobryant 7y ago

> it never quite came through. Hard disagree. The Clojure ecosystem is great.

downerending 7y ago
It's obscure enough that it perhaps proves my point. Nonetheless, I'll have to look at it.
- bcrosby95 7y ago
  If you haven't heard of Clojure, you probably haven't listened to a Rich Hickey talk. They're worth it even if you don't use Clojure. "The Value of Values" and "Simple Made Easy" in particular.

kazinator 7y ago

The division was never a problem during the heyday. The decline in Lisp in fact started at the height of standardization. > it's so easy to write a lisp system Have you tried it, and how far did you get? Making a Lisp system useful for production is pretty brutal. 13 years into it you will be debugging some eleven-year-old GC or compiler bug. People think it's easy because they visualize the job as being done when some executable file converts (+ 1 1) into 2.

downerending 7y ago
Maybe it declined for other reasons, but my impression was that the Lisp world was very splintered. I certainly can't think of any other example of a set of quite similar languages that's anywhere near as large. The Xerox Dandelion and friends was a thing to behold. If you ever got to touch one, you never forgot it. In my school, implementing a basic Lisp interpreter was part of a required class in the CS curriculum. (If it's not still, everywhere, well, then for shame.) Yeah, production is something else entirely. By "easy", I mean that dozens if not hundreds of people were tempted to write their own slightly better Lisp. Many of those did reach a production-ish level. But then what? There was no flag to rally around--just lots and lots of niche systems. I'm not sure. I think if the community had rallied around Common Lisp or Scheme, or maybe just those two, it might have ruled the world. It just didn't happen. And I'm miserably sad about that. What do we have now? Elisp is great, but I can't build production on emacs. I'll look at Clojure, but I'm dubious. Python is kind of lispy, but has had its own destructive schism. And C++ marches on--somewhere buried in there is a mildly functional lisp, using the worst of all possible syntaxes.
- TeMPOraL 7y ago
  I'd think a refreshed Common Lisp would still have a chance. It's a great language with some really solid open-source implementations, but it got standardized at the transition point in our industry, so the spec doesn't even consider (now-commonplace) things like threading or networking, while exposing you to some abstractions over systems died out. The problems of CL wouldn't be unsurmountable if the community was larger, though. All the important stuff that wasn't standardized gets added by each implementation anyway, and then portability libraries get created that ensure consistent interface. But the state of most libraries is... rough at the edges. I contrast that with Clojure, with which I spent some months over the last two years. The library ecosystem (not Java-side, but Clojure-specific) is great, and a lot of care goes into it. I particularly remember being in awe of just how thorough Liberator is[0]. -- [0] - http://clojure-liberator.github.io/liberator/tutorial/decision-graph.html http://clojure-liberator.github.io/liberator/tutorial/decisi...
- lispm 7y ago
  > I certainly can't think of any other example of a set of quite similar languages that's anywhere near as large. Scripting languages on Unix.