Tools for building a Cobol compiler

Robert Dewar <dewar@cs.nyu.edu> wrote:
>THe kind of problems that Vadim cites in using YACC to compile COBOL are
>typical of the steps you go through making a YACC grammar for any real
>language. COBOL is neither particularly more difficult, not particularly
>more easy than other languages to parse using tools of this type. Similarly
>it is simple enough to use Lexx to do the lexical analysis.
>
>Why of course anyone in this day and age would use such antiquated tools
>as Yacc and Lex to implement anything approaching a real compiler is beyond
>me. It makes sense as a student excercise, but if you are seriously building
>a compiler, there are far better tools available (and of course the lexer
>and parser are entirely trivial parts of the compiler in any case).


Well, I have some reasons:


- YACC is verified, reliable, and it's everywhere.
- YACC never was slow, or took enormous amount of memory.
- YACC is free.
- I know YACC.


Speaking of LEX, I prefer not to use it in a serious project. It tends to
build huge tables, be somewhat slow, and more importantly, many things
cannot be expressed in LEX easily.


However, we all understand that these pro-YACC arguments are somewhat
weak, since a lot more expressive power is neede to parse the real
languages like COBOL. So I would like to ask you guys: what are these 'far
better tools'? And where can I get them?


I guess GMD cocktail will be mentioned. While I heard of several such
tools, it's always difficult to tell from announcement/promo if the tool
works and if it has enough power. So it would be good to hear about
real-life experiences with such tools.


Vadim Maslov
--

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