4 ms·
> So what is the problem with creating domain-specific languages as a problem solving technique? The results are very efficient. [...] It results in many sub-la
by karmakaze 7y ago
> So what is the problem with creating domain-specific languages as a problem solving technique? The results are very efficient. [...] It results in many sub-languages all slightly different. This is the true reason why Lisp code is unreadable to others.
> The reason Lisp failed was because it fragmented, and it fragmented because that was the nature of the language and its domain-specific solution style.
The main points above can be solved with two things:
1. A more opinionated base language with batteries (e.g. Clojure)
2. A package manager shared by users of the language
Ruby is expressive and has the DSL failure modes as lisp (though usually not as low-level), but it starts with a more batteries-included base. If it wasn't for Rails and gems leading the way to normalized usage Ruby would not be as successful as it is today.
- janoc 7y agoSorry but that's just wrong. He is complaining about DSLs making things hard to read. That's not fixed by switching to another language or a package manager. That's purely a software engineering problem - nobody forces you to write code in that style. And LISP code being unreadable to others - sorry, that's bull. Seen e.g. some "modern" Javascript stuff recently? Good luck making heads and tails out of some of the frameworks - it often doesn't even look like a Javascript anymore! And nobody seems to be claiming that it leads to the demise of the language. Or numerical calculations in Fortran - still a gold standard for scientific stuff. To me the article is very much an output of someone who couldn't have been bothered to learn the language griping about things he doesn't really understand well and making huge generalizations based on that. Why LISP isn't popular today is simply because that unlike for other languages, there hasn't been a free/cheap compiler and IDE for a PC for a very long time (except for the crippled LispWorks). LISP has always been an university/research thing running on either on specialized machines or later on Unix, not something "mere mortals" had access to. Also most programmers have been taught languages like Pascal/C/Java, maybe Scheme in their introductory courses and have never been exposed to LISP, so they have no way to know about it unless they are themselves curious about it.
- karmakaze 7y ago> Sorry but that's just wrong. He is complaining about DSLs making things hard to read. That's not fixed by switching to another language or a package manager. I wouldn't put the formatting of the adhoc dsls used in a program as the hard part. It's the chosen abstractions and decomposition/composition patterns used. This is true in any language with metaprogramming but why has Lisp suffered more than others? I doubt reader macros. If anything they would help put things in logical rather than implementation required form.
- dreamcompiler 7y ago> there hasn't been a free/cheap compiler and IDE for a PC for a very long time CCL is a free Common Lisp compiler that runs on Mac, Windows, and Linux. On the Mac, it comes with an IDE. It's been around for about 20 years. SBCL is another free compiler that uses Emacs/Slime as an IDE.
- AgentOrange1234 7y agoI agree that CCL and SBCL are superb on Linux and Mac. But until at least a few years ago, they were both poorly supported on Windows (requiring something like Cygwin). That may have changed now. Also, AFAIK, CCL is dying now. :(
- fiddlerwoaroof 7y agoCCL's development is relatively slow compared to sbcl, but I'm not sure that it's "dying". And, there's been a brand new compiler on LLVM, clasp, as well as the SICL project.
- flavio81 7y ago>AFAIK, CCL is dying now CCL is the 2nd most downloaded Lisp implementation, so it can't be called "dying." You'll find that new lisp projects are almost invariably tested on CCL too.
- AgentOrange1234 7y ago