Re: Why separate Lexical & Parser Generators

John Heron <heronj@smtplink.NGC.COM> writes:
  > Pardon me if this question is naive. Why have a separate parser generator
  > and lexical analyzer generator?
[...]


In addition to our moderator's comments one might add:


We have lots of implicit shift/reduced and sometimes reduce/reduce
conflicts in the regular portion of the grammar that corresponds to the
longest match and first match rules in Lex. Also, many languages, like C,
have difficult constructions (typedef:ed names in C) that require the
lexical analyzer to do smart things. In a common grammar this corresponds
to some form of conditional parsing rules, so we must add stuff like that.


There have been some research into lexical analyzer free parsing. For
example:
Salomon, D.J. and Cormack, G.V. Scannerless NSLR(1) Parsing of Programming
Languages, ACM SIGPLAN Conference on Programming Language Design and
Implementation, June 21-23, 1989.


Anders Andersson
andand@csd.uu.se
--

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