Self-Referential Language?

[alexander]

you dont need to know assembly or c to write your language, you can write it in any language you want, generally if it's a compiled language then the compiler is written to smartly translate the language into assembly, and then it is assembled and compiled again into a binary for a particular operating system...

Also most interpreters, for the sheer need of speed of execution, are written in C (more oftenly C++), but as Buffy could tell you, you can write a C compiler in Lisp, nothing's stopping you. Mind you most C compilers are written in C ( yes, go figure that one ) also most (well i only know of one, and a shell) C interpreters are written in C.

Point being, no, you dont need to know a low level language to make your own programming language. Infact a well-defined language would tend to attract programmers that will code it in a low-level language for you. There are a few books available on language design

David A. Watt - "Programming Language Design Concepts"
Raphael Finkel - "Advanced Programming Language Design"
Terrence W. Pratt and Marvin V. Zelkowitz - "Programming Languages: Design and Implementation"
Bruce J. MacLennan - "Principles of Programming Languages"

At least a couple of those above should be augmented with "Programming Language Pragmatics Second Edition" by Michael L. Scott

Also these three books in succession (they get progressively more hardcore, and they all talk about modern compiler techniques, loop optimization, SSA, all kinds of good stuff):
Kenneth C. Louden - "Compiler Construction: Principles and Practice" - (this is more of a course book for students)
Andrew W. Appel - "Modern Compiler Implementation in C" - (this is a book that those students read to ace the course and familiarize themselves with up-to-date practices)
Steven Muchnick - "Advanced Compiler Design and Implementation" - (this is the book for the students that can teach the course and have the professor learn things he's probably never heard before)

But also remember, if you go with my advise and get 5 books (one out of the first set, the pragmatics book and the last 3 on compiler design) its nearly a 500 dollar bill, books on advanced topics are not cheap... But my point is, programming language design is above the range of a tutorial, a good programming language takes knowledge and practice, psychology and imagination, a good knowledge of current languages, but not a drive to compete with any of them. Only then can you desing a language that stands out on its own...

[alexander]

Also remember that most of the greatest programming languages were written for the designer and his friends, most bad languages were also written for that purpose, the best language IMHO, Python, was written with a specific mission in mind, but there are plenty of really bad languages out there that were written for that purpose too, so design it to use for yourself and your friends with a mission, hopefully you'll do better then the thousands of horrible, failed languages out there

[Pyrotex]

Yes, that was it: BNF. Backus Naur Format. Thanks Craig.

KAC: The logical start IMHO would be defining the starting point BNF in XML, then writing an Interpreter in Python that used the BNF explicitly as the definition of the language itself. Now, for the starter language, I would begin with a simple subset of Python!!! I would only add one teensy feature to this subset of Python, something like a keyword, say "BNF(n,m)", where n is the nth statement in your BNF, and m is the mth option in that statement. (The options are the text between the vertical bars |.) The type of BNF(n,m) would be a string, of course.

[TheFaithfulStone]

Originally Posted by KAC

Also, Does anyone have links to resources related to writing languages? Tutorials, examples, compiler/interpreter/profiler examinations, and the likes? Do I need to learn assembly and/or C to write my own language?

Not at all. I've written several DSLs in Python using ANTLR as a parser generator.

It's not really what I'd call fun but it's not too bad.

I also wrote one in Perl and generated the tokenizer and parser by hand - and THAT was my own special version of hell.

tfs

[alexander]

perl is the perfect langauge for writing a parser actually, cuz the one thing it does really well is work with text

[KickAssClown]

It seems I maybe moving onto learning a new notation set, vW-Grammar.

Alright, I've determined that I we need to discuss Abstract Syntax (AST), and Parsing Trees. Also, I feel we might need to discuss abstract semantic graphs. I'm aware that these structures can be utilized to facilitate reflection within a language, but I'm unaware of how one would do that.

Does Java use it's AST to provided reflective capabilities?

[KickAssClown]

Alright, so I've been spending a lot of time lurking around the Semantic Web Interest Group Channel on the Freenode IRC network. We've been talking about defining an ontology for describing programming languages in a data-language format. one of the people there, Toby Inkster, has gone and started an implementation which is available.
core terms
one particular view of how programming languages relate together
one particular view of how data languaes fit together.

This schema is purposed for use as the declarative portion of a Language of Languages.

The basic idea which we arrived on is that language is practically impossible to define manually, but we have various implemented languages that we can parse programmatically to build towards an implicit definition of language.

We arrived at this conclusion based on the fact that a full vW-grammar is Turing complete and therefore undecidable as a metalanguage. It will result in an infinite number of languages. If a universal language exists, it would have to express the set of all expressible languages, so any language known of would likely be a subset of a Language of Languages.

enjoy the links,
Ian Mclean