Learning Lisp - Why? | Hacker News Reader

Learning Lisp - Why?

stackoverflow.com

14 pointsecounysis28 comments

I learned Common Lisp to see what all the fuss was about, and I don't regret it. I think most claims in its favor are hyperbole, but the Lisp implementations are very nice pieces of technology, providing features comparable to CPython with approximately the performance of Java. I almost always find myself missing some feature of the Common Lisp environment when working on more lucrative endeavors.

"Learning Lisp is for the programmer what meditation is for the Buddhist monk ..."

http://stackoverflow.com/questions/4724/learning-lisp-why/27...

Stack Overflow is down for maintenance, so here's a cached version:

http://webcache.googleusercontent.com/search?q=cache:QfU7Ls3...

"Lisp is worth learning for the profound enlightenment experience you will have when you finally get it; that experience will make you a better programmer for the rest of your days, even if you never actually use Lisp itself a lot."

- Eric S. Raymond, "How to Become a Hacker"

http://www.paulgraham.com/avg.html

Has there been ANY programs written in lisp that has risen to the status of anything more than novelty?

IBM's watson, the best thing we got going for artificial intelligence, is it written in lisp? NO, it's written in java. The link from lisp to artificial intelligence is a lie. Lisp is just a very poor language, it's designed to get eager novices working on things that are irrelevant so they will learn (the hard way) that being a good programmer means laughing at the people who promote lisp as "an epic language". Lisp had its day, somewhere back in the 1950's/1960's. That day is gone.

If I'm working on any sort of problem needing a programming solution, and someone suggests we use lisp, I buy that person a copy of "The pragmatic Programmer", and tell them to read it closely.

emacs is written in lisp (and some c for the interpreter), and it must be considered more than a novelty.

And autocad uses its own version of lisp (autolisp) as a scripting language.

Edit:

There are also the Lisp Machines from the 80's (a bit old, yes, I know) [1].

And the Jak and Daxter series of games[2]

[1] - http://en.wikipedia.org/wiki/Lisp_machines

[2] - https://secure.wikimedia.org/wikipedia/en/wiki/Game_Oriented...

Viaweb?

Clojure is going to provide the true test to see if CL and Scheme's failure to take over the world is due to inherent problems with lisp or just inertia due to lack of 3rd party library support.

With startups like BackType and FlightCaster getting acquired, it's a positive sign.

I'm looking forward to seeing how Clojure improves with invokedynamic in Java 7, and lambdas and closures in Java 8. Hope the ramp to maturity for a Java 8-based Clojure is a quick one.

There have been Common Lisp and Scheme implementations in the JVM for ages (Kawa, ABCL, and many others -- I wrote one myself called Skij). It's not clear to me why Clojure took off when those others didn't. But if reinventing Lisp is what it takes for Lisp to get traction, I'm for it despite my CL roots.

>Has[ve] there been ANY programs written in lisp that has risen to the status of anything more than novelty?

Yes.

Has there been ANY programs written in lisp that has risen to the status of anything more than novelty?

The program you used to post that question is written in Lisp. HN probably counts as more than a novelty by this point.

Surely most of the software running this site is simply HTML with some basic data movement and storage going on in the backend.

I'm not sure this really even qualifies as 'software' ;)

It could have been written in php/asp/python/perl etc and would have functioned exactly the same, be about the same lines of code, as maintainable, as prone to bugs and so on.

In fact, I seem to remember Reddit rewrote from lisp to python in a weekend?...

If you haven't looked the source, and therefore don't know what it does or how long it is, how can you know whether it would be the same length translated into other languages?

He can't, so he is just making an assumption (whether from experience or not, who knows). But I would be interested in some information about the HN codebase...

Is it shorter? Less error-prone? Simpler?

https://github.com/nex3/arc/blob/master/news.arc

Lines of code is probably the wrong term for me to use. Since it's bogus. (Depends on how much the language provides, libraries, etc and doesn't necessarily capture program complexity).

A more accurate term would be "Required programming effort".

Would you agree that it would take a similar amount of effort for a proficient PHP programmer to write HN, as a proficient Lisp programmer etc?

FWIW, I have just looked at the source and IMHO I'd expect it to be extremely similar in PHP/other languages.

It certainly qualifies as "software." What possible definition of "software" would exclude a web application?

Obviously, literally any software could be identically implemented in any language. Lines of code don't really matter, but I strongly expect that the Arc code of HN is much more concise than an implementation in php/asp/python/perl would be, even if you used a web framework. PG prides himself in the conciseness of the HN code, and Arc seems to lend itself to concise implementations of web apps.

As for maintainability and bug proneness, both depend more on the programmer(s) than the language.

> It certainly qualifies as "software." What possible definition of "software" would exclude a web application?

My point was that it doesn't do heavy algorithm lifting, processing etc. It's not a phong shading ray tracer or anything.

It's moving data around, storing it and displaying it. It's mostly plumbing work - getting data from one place to another.

I'm sure I could write a simple phong shading ray tracer faster than I could build a copy of this web site. Re-implementing a well known and well understood algorithm is easy, making a working and popular application that people enjoy using is hard

In pg's ANSI Common Lisp, he defines a ray tracer in one page of code, creates all the necessary functions for spheres in another page, then inserts 28 spheres and prints out the world in a dozen lines of code. There's part of a page of utility functions, like creating a point structure. You can examine the code here: http://lib.store.yahoo.net/lib/paulgraham/acl2.lisp . Search for the first instance of "surface".

I don't know anything about ray tracers, but if it's an algorithm, then it's "mostly plumbing work" in the same way you say a web application is. The beauty of the world we live in is that you can compute anything that can be computed with an absurdly small set of basic operations.

Actually, there's very little HTML in the news.arc codebase. The front-end bits are mostly functions that generate HTML in a rather abstract way.

Also, at a certain level of skill, you can rewrite just anything in a weekend — but that doesn't mean the code is as clean, as lean, as maintainable or as correct. It just means you've approximated the most obvious functionality in the software.

Just FWIW, This is still mostly HTML:

  (trtd ,@body)
         (trtd (vspace 10)
               (color-stripe (main-color ,gu))
               (br)
               (center
                 (hook 'longfoot)

This is HTML, and yet this is Lisp. That's an example of Lisp powers in use.

"If I'm working on any sort of problem needing a programming solution, and someone suggests we use lisp, I buy that person a copy of "The pragmatic Programmer", and tell them to read it closely."

Then I'd recommend you go back and read it closely yourself, because there's an entire chapter on domain-specific programming which includes tips on implementing mini-languages to write code close to the problem space, which is a very Lispy suggestion.

Doesn't ITA count? (Powers Orbitz, acquired by Google for a fairly significant amount of money).

Also see here: http://www.franz.com/success/

"lie", "things that are irrelevant" ... Opinionated words. I have not read the pragmatic programmer, but I fear that you just gave it a bad rap.

Reply on news.ycombinator.com