Hacker News Reader

Comment by baxter | Hacker News Reader

I agree that learning how to create a programming language is a valuable lesson in and of itself -- I may even buy the book to see what it's like -- but claiming that you can create a "revolutionary" programming language within a few days conveys the wrong message.

Creating programming languages is not difficult but designing good programming languages is.

View on HN

I am designing a (hopefully) revolutionary programming language. (At least, revolutionary in its target domain; a general purpose DSL, if you will).

To do this, I am not reading a 44 page PDF to tell me everything I need to know, instead I'm reading thousands of pages of material(1), I've been prototyping code, talking to people and reading a lot of Lambda the Ultimate discussions. You just can't learn all the fundamental concepts, all the theory and the practical details (language and implementation/compiler/vm) from 44 pages...

1: Some of the stuff I've been reading especially for this: Concepts, Techniques and Models of Computer Programming - 930 pages of fundamental programming language theory. Structure and Interpretation of Computer Programs - 657 pages of Scheme goodness. Advanced Topics in Dataflow Computing and Multithreading - 460 pages of research. Countless SIGPLAN papers. God knows how many web pages and random PDF's.

And thats just the theory, I've yet to get reading the actual COMPILER books. Near the top of my list is the dragon book - another 1000 pages.

Creating a new programming language is trivial. I just invented one called "awesomepython". It's just like python, except that the keyword "and" is replaced by "awesomeand", "if" is replaced with "awesomeif", and so forth. The compiler is one line of awk.

Creating a language that's actually good for something is a very different sort of challenge...

"Creating a new programming language is trivial. I just invented one called "awesomepython". It's just like python, except that the keyword "and" is replaced by "awesomeand", "if" is replaced with "awesomeif", and so forth. The compiler is one line of awk."

Coffee all over keyboard! Damn you!! :-D

Hilarious! Upvoted! Now to clean that spilt coffee!

yeah the marketing is too traditional!, but hey it didn't mention "become the next programming rockstar" :-)

But I mean, if it is a walkthrough then for people interested in the JVM it could be useful.

I'm buying it because I found clear, comprehensive literature explaining the fundamentals of the JVM severely lacking (suggestions welcome, however!).

You can start by playing with the ASM library. It's a java library to parse, generate and analyze JVM bytecode. Here's the official guide: http://download.forge.objectweb.org/asm/asm-guide.pdf

While learning how to use ASM you will also learn about the instruction set of the JVM. It also includes a disassembler that, given a .class file, outputs the java code necessary to generate that class using ASM.

Wow, thanks! You'd be amazed how long I spent looking for the advice which probably took you a couple of minutes to type.

That's an interesting observation, touching on the education discussion - that the bottleneck in advancing individual learning/human knowledge questions is still that of finding someone who knows the answer. Universities provide one solution by creating a "critical mass" of smart people nearby that you can go to for advice. And although maybe my academic network is lacking, I couldn't quite get the same quality of advice reaching out personally as what you gave. StackOverflow and news.YC help [1], but didn't quite give me the solution I was looking for either (until now).

If I was convinced I could consistently good advice from a service, I'd pay well for it - upwards of a thousand dollars a year, I'd imagine.

1. http://news.ycombinator.com/item?id=451655

Reply on news.ycombinator.com