Related articles |
---|
Lazy Evaluation of a DFA eborduas@microstar.com (1995-11-23) |
Re: Lazy Evaluation of a DFA Martin.Jourdan@inria.fr (1995-11-28) |
Re: Lazy Evaluation of a DFA oz@nexus.yorku.ca (1995-11-30) |
Re: Lazy Evaluation of a DFA kanze@lts.sel.alcatel.de (1995-12-09) |
From: | kanze@lts.sel.alcatel.de (James Kanze US/ESC 60/3/141 #40763) |
Newsgroups: | comp.compilers |
Date: | 9 Dec 1995 19:38:51 -0500 |
Organization: | SEL |
References: | 95-11-205 95-11-254 95-11-261 |
Keywords: | lex, performance, DFA |
oz@nexus.yorku.ca (ozan s. yigit) writes:
|> I wrote:
|> I think this is wrong. common egrep builds a complete dfa for speed.
|> Well, Don Pardo tells me he knows a common egrep that uses lazy
|> construction. I am almost sure this cannot be SYSV egrep. :)
I'm going back a long time, now, but I seem to remember early Unix
manuals stating that egrep was very fast on large input, but had
significant start-up time. Somewhere along the line, this start-up
time disappeared. Could this be due to a change from building a
complete DFA to using lazy evaluation? (On all but very large input,
lazy evaluation will be a win, because of the number of states that
never get evaluated.)
--
James Kanze Tel.: (+33) 88 14 49 00 email: kanze@gabi-soft.fr
GABI Software, Sarl., 8 rue des Francs-Bourgeois, F-67000 Strasbourg, France
Conseils, Etudes et realisations en logiciel oriente objet --
-- A la recherche d'une activite dans une region francophone
--
Return to the
comp.compilers page.
Search the
comp.compilers archives again.