Re: Speeding up LEX scanning times

dimock@das.harvard.edu (Allyn Dimock)
Thu, 2 Feb 1995 14:44:57 GMT

          From comp.compilers
Related articles
Speeding up LEX scanning times pahint@eunet.be (1995-02-02)
Re: Speeding up LEX scanning times dimock@das.harvard.edu (1995-02-02)
Re: Speeding up LEX scanning times c1veeru@WATSON.IBM.COM (Virendra K. Mehta) (1995-02-02)
Re: Speeding up LEX scanning times monnier@di.epfl.ch (Stefan Monnier) (1995-02-03)
Re: Speeding up LEX scanning times mercier@hollywood.cinenet.net (1995-02-03)
Re: Speeding up LEX scanning times vern@daffy.ee.lbl.gov (1995-02-04)
Re: Speeding up LEX scanning times eifrig@beanworld.cs.jhu.edu (1995-02-07)
| List of all articles for this month |
Newsgroups: comp.compilers
From: dimock@das.harvard.edu (Allyn Dimock)
Keywords: lex, performance
Organization: Aiken Computation Lab, Harvard University
References: 95-02-010
Date: Thu, 2 Feb 1995 14:44:57 GMT

It is my impresion that any simple compiler is going to spend a
substantial percentage of its time in the lexical analysis phase: Lets
face it, (readable) source text is not a very compact representation, and
the lexer has to make several decisions per character.


Your 80% sounds high, the number that was quoted to me as standard was
">25%" Many versions of the utility lex allow you a choice of several
algorithms trading off table space vs time. You might try requesting a
different algorithm. It is my impression that flex gives you quite a
number of algorithms to choose from.


--
Allyn Dimock dimock@das.harvard.edu
Aiken Comp. Lab. #111 w (617) 495-3998
33 Oxford St. h (617) 451-4565
Cambridge, Ma. 02138 249 A St. #63 / Boston, Ma. 02210
--

Post a followup to this message

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