LispCast - Latest Comments

Re: PEG Parser for Clojure

Freshila — Fri, 04 Feb 2011 04:19:39 -0000

hello .
nice post .

Re: Content scraper (based on Readability) in Python

sari — Mon, 24 Jan 2011 03:38:20 -0000

gretings .
nice post .
thanks .

Re: Haskell syntax vs. Lisp syntax

Eric Normand — Sat, 27 Nov 2010 02:46:59 -0000

Thanks for the help.

Re: Haskell syntax vs. Lisp syntax

Guest — Fri, 26 Nov 2010 23:54:30 -0000

First of all, there's the GHC Libraries page which describes most things, if you know which module they're in (http://www.haskell.org/ghc/...

As for a Google alternative, there's the aptly-named Hoogle (http://haskell.org/hoogle/) which searches fine for any Haskell function, and you can also search by type signature.

Re: Constraints

Mimo — Fri, 26 Nov 2010 17:27:38 -0000

This articl very gooooood . I like it >

what is the constraint that occur as a result of external circumstances , Ex?

Re: Content scraper (based on Readability) in Python

Paulus — Sat, 20 Nov 2010 11:47:49 -0000

For an earlier attempt, see https://github.com/gfxmonk/.... It has performane issue for me, though - it can sometimes take seconds to parse reasonably long document.

Re: PEG Parser for Clojure

Eric Normand — Tue, 09 Nov 2010 07:06:54 -0000

My library aims to make a DSL for writing grammars. amotoen appears functional (I haven't used it) but the grammars written for it do not look very readable to me. I would like to write grammars very easily and readably.

Note: clj-peg (my library) is currently far from that goal.

Re: Google likes SEO

Shane Hoversten — Mon, 01 Nov 2010 16:59:52 -0000

This is a good post, but I also think it's important to keep in the forefront of your mind that using whatever blessed Google metadata or whatever isn't to help Google; it's to help you. If it were to help Google, surely nobody would bother; the fact that anyone would bother proves that there's rewards to be had, and people with websites want those rewards. Along the way you benefit and Google benefits, and in fact any other search engine that wants to use that data benefits, too.

Of course, you're right in pointing out the dark side of it all. Invention of a system is, implicitly, an invention of techniques to game that system. I trust you are right that a lot of unworthy sites get bumped higher than they ought to; but that's always been true, back when Google's analysis was purely static link structures. I have not noticed search results to be worse recently; although I suppose I haven't noticed them to be better, either.

Broadly speaking, I've been waiting for something like this for a long time. To do search at the level of intent -- returning results that reflect what people really want when they use a search engine -- basically requires strong AI to be solved properly; link analysis is a bastard trick that happens to do a pretty good job. I've always thought it a shame that the contents of my site would have to be inferred by search engines using the algorithmic equivalent of a blind man stumbling around holding his arms in front of him. How much better, I always thought, if I could use my own strong AI caliber brain to be explicit about it. Again, I think your concerns are legit; but I'd like to give this kind of explicit semantic hinting a chance.

Re: PEG Parser for Clojure

edbond — Sun, 24 Oct 2010 16:05:36 -0000

Did you see http://github.com/richard-l... ?
Can you compare it to your lib?

Re: Framework of Programming Language Expressivity

Eric Normand — Wed, 20 Oct 2010 01:21:29 -0000

Thanks for your comment.

You say that I assume that Java developers don't know how to design their software. I am not assuming that. What I am assuming, for that specific, abstract example, is that the features of the software could not be abstracted faster than linearly for the LOC metric in Java.

What that means is that for every feature added, some nearly constant number of lines of code has to be added. For instance, if every feature needed a new class to be added to the project, just the line "public class . . ." would be sufficient to keep it linear.

I think what you describe is accurate: Java projects create a single level of abstraction, namely the domain model, then write lots of code to manipulate objects in that model.

Scheme (and other Lisps, and Smalltalk, and probably others) quickly bootstrap away not only the domain model, but also the logic that in Java is typically written in Java code, into some form of declarative DSL that is much more concise, readable, and extensible. Doing this in Java is possible but is not currently considered a best practice.

Re: How to write software

Eric Normand — Tue, 28 Sep 2010 11:50:55 -0000

Yeah, I've gotten a lot of inspiration from Moore. Especially the "no abstraction" idea.

Re: Framework of Programming Language Expressivity

A_flj_ — Sun, 15 Aug 2010 16:31:33 -0000

Idunno. It seems to me that your assumption that beyond a certain functional complexity Scheme allows you to write smaller programs providing the same functionality is based on the assumption that Java programmers don't properly design their code. This assumption, IMO, stems from you not knowing how Java projects are usually performed.

Java programmers do indeed lack a mechanism of macros, but nobody prevents them from designing highly reusable components/classes, which should make development of additional features easy. In fact, it's what most projects do: develop a relatively small and simple set of domain-specific classes, then build huge applications on top of the application-specific library. In fact, you can even include rhino or beanshell as a library in your app and drive your app from scripts.

Therefore I'd very much like to see a real life example before I g ive any credit to your assumptions.

Re: A Killer App for Common Lisp

www.starcol.org — Wed, 04 Aug 2010 07:05:48 -0000

very good

Re: How to write software

Samuel A. Falvo II — Tue, 27 Jul 2010 15:50:14 -0000

Eric, the model you disclose here overlaps much of the "thoughtful programming" practice encouraged by Chuck Moore (who invented the Forth language). :)

Re: How to write software

tour ke china — Tue, 20 Jul 2010 05:14:55 -0000

thanks for the info and explanation provided

Re: Why type systems are wrong

sewa mobil — Sat, 26 Jun 2010 00:32:03 -0000

nice description about type systems
www.crownrentcar.net

Re: A Killer App for Common Lisp

Issacmartin20 — Thu, 24 Jun 2010 04:46:05 -0000

Before reasing the article Paul Graham was totally unknown to me.

Re: A Killer App for Common Lisp

Issacmartin20 — Wed, 09 Jun 2010 23:19:07 -0000

Thank you very much for lots of valuable information. Issac

Re: How to write software

Eric Normand — Wed, 09 Jun 2010 13:56:19 -0000

You're welcome.

Re: How to write software

Eric Normand — Wed, 09 Jun 2010 13:56:11 -0000

Thanks

Re: How to write software

z... — Sun, 30 May 2010 05:53:29 -0000

surprisingly coherent

Re: The next LLL killer: High Level Optimization

Jonathan J — Sun, 30 May 2010 05:31:56 -0000

Common Lisp has compiler macros that could do the (length= 10 ...) optimisation easily. :)

Re: How to write software

jorge mendes — Sat, 29 May 2010 03:19:01 -0000

Thanks for a very systematic, simple and clean process. It will surely hel me.

Re: Writing map in Haskell

sewa mobil — Wed, 26 May 2010 03:00:41 -0000

Hi I saw your blog. You have done a good job, I really liked your blog and very informative.

Re: A Killer App for Common Lisp

Jonsulred — Fri, 07 May 2010 09:00:31 -0000

Someone should set up a community site with the Dolphin cms. That would be awesome to have a social site devoted to lisp programmers!
I was thinking of started one, but just got a new baby so gotta crunch for work :/