Re: Help needed on LALR(1) ambiguity

philip.k.chow@gmail.com
Tue, 18 Nov 2008 16:46:02 -0800 (PST)

          From comp.compilers
Related articles
Help needed on LALR(1) ambiguity philip.k.chow@gmail.com (2008-11-14)
Re: Help needed on LALR(1) ambiguity cdodd@acm.org (Chris Dodd) (2008-11-17)
Re: Help needed on LALR(1) ambiguity armelasselin@hotmail.com (Armel) (2008-11-17)
Re: Help needed on LALR(1) ambiguity svenolof.nystrom-nospam@bredband.net (Sven-Olof Nystrom) (2008-11-18)
Re: Help needed on LALR(1) ambiguity philip.k.chow@gmail.com (2008-11-18)
Re: Help needed on LALR(1) ambiguity Danny.Dube@ift.ulaval.ca (2008-12-01)
| List of all articles for this month |
From: philip.k.chow@gmail.com
Newsgroups: comp.compilers
Date: Tue, 18 Nov 2008 16:46:02 -0800 (PST)
Organization: Compilers Central
References: 08-11-055 08-11-083
Keywords: parse, LALR
Posted-Date: 19 Nov 2008 19:50:24 EST

Hi everybody, thank you for your comments.


Chris Dodd wrote:
> So the problem is that you need to be able to look
> ahead to the end of the input to see how many |-suffixes
> there are in order to figure out how to parse the beginning
> of the expression. This makes it pretty much impossible
> to parse without unbounded lookahead, or backtracking,
> or something else.


Yes I believe you're right.


Sven-Olof Nystrom wrote:
> In your example, making the second production of "Names" left-recursive
> makes the grammar LALR(1). Yacc detects no conflicts for the following
> grammar:
>
> names : ID COLON
> names : names COMMA ID


Unfortunately this is not the same language.


My names match the following:
a:
a,b:
a,b,c:


Your names unfortunately match the following:
a:
a:,b
a:,b,c


I understand now that there is probably no way to solve my problem
without unlimited lookahead. Thank you everyone for your help.


Post a followup to this message

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