Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!mailrus!ames!lll-lcc!pyramid!infmx!andyk
From: andyk@infmx.UUCP (Andy Kashyap)
Newsgroups: comp.lang.c++
Subject: Re: C++ considered disenchanting
Message-ID: <572@infmx.UUCP>
Date: 27 Oct 88 07:19:51 GMT
References: <441@grand.UUCP> <1000004@hpclscu.HP.COM> <4634@polya.Stanford.EDU>
Reply-To: andyk@infmx.UUCP (Andy Kashyap)
Organization: Informix, Menlo Park, Ca. U.S.A.
Lines: 19

In article <4634@polya.Stanford.EDU> shap@polya.Stanford.EDU (Jonathan S. Shapiro) writes:
 - In article <1000004@hpclscu.HP.COM> shankar@hpclscu.HP.COM (Shankar Unni) writes:
 - >You don't necessarily need an infinite-lookahead lexical analyzer or
 - >heuristics or anything, if you can do one thing in the lexer: recognize 
 - >typedef names IMMEDIATELY, and use a different token for typedef names.
 - 
 - Unfortunately, there are places where you cannot correctly determine
 - that a name should be introduced into the typedef table without
 - infinite lookahead.
 - 
 - Jon

Furthurmore, if you try to immediately recognize a token in lexer, you
also immediately lose the context ... unless you want to pass lex the
context; but that's what the parser does.

Hmmm, did I just paraphrase Shapiro??

- andy kashyap @ Informix (415) 322-4100 x685