Compiler writers will love this language

I think that some languages would be a lot easier to implement and or
better designed if experienced compiler writers were there right from
the beginning when Mr/Mrs Language Designer were designing their new
language. If Mr/Mrs LD decided to have feature A, the compiler writer
would say "oh no, that would affect the grammar in such and such a way
- which is not fun to implement". If Mr/Mrs LD decided to have
semantics B, the compiler writer would say "that's no problemo, bring
it on".... After lengthy negotiations a cool language is produced and
its reasonably easy to implement.


So here we are. Imagine for a moment that I am a language designer and
*you* are my faithful compiler writer. I am all excited about a
language X that I am going to design and many ideas are coming my way.


However seeing that I am a polite and considerate language designer, I
am going to consult you and seek your advice on certain aspects of X,
so as to make your life easier when it comes to implementation. While
I may not end up granting all your requests, am still interested in
hearing them.


So then, according to your experience, what are your guidelines? The
kind of answers am looking for are in this form:


*if you decide to have _____ then make sure you have/dont have ____
due to the way they interact when mixed together


* you can include feature _____, its useful and easy to implement


* keep in mind that ____ is harder to implement than ____ but they do
the same/similar thing


* stay away from ____ its not worth the trouble unless you need ____


* look at the grammar for language _____ , its easier/harder to
implement because it does/doesn't have _____


* grammars which have/dont have _____ or are like ____ are good/bad
because I can/can't do ____


* ensure your semantics for feature A are like ____ because of _____


* language ___ does ____ in this way, but it would be better to do it
like ____ because then we would implement it using _____


* _____ is better kept in the language because then I can do ____ to
optimize


* but _____ should go in libraries rather than the language because of
_____




you know, that sort of thing. You don't have to fill in the blanks to
the *exact* sentences above - feel free to answer the way you see fit.
I'd appreciate it if you could:


* give examples of languages I can refer to with respect to your
recommendations


* state which languages you've been involved in implementing and in
which year


* give reasons for your recommendations where possible


* point me to an papers/books that deal with what affects the
implementability of programming languages


* address any or all things in the categories: syntax, semantics,
code generation, optimization, memory management, run-time
environments (and whatever else you want)


It's like having an opportunity to create a language that you'll
*love* to implement, so fire away! Many thanks in advance :)


Cheers, E.


Ps: I am interested in programming language design and compiler
writing. I know more about the former than the latter, hence this
post. I am also reading the classic text: "Compilers: Principles,
Tools and Techniques" by Aho, Sethi and Ullman.

Return to the comp.compilers page.
Search the comp.compilers archives again.