Re: Purple Dragon Book: Newbie questions on Chapter 4 text.

Harry <simonsharry@gmail.com>
Thu, 10 Jun 2010 21:21:30 -0700 (PDT)

          From comp.compilers

Related articles
Purple Dragon Book: Newbie questions on Chapter 4 text. simonsharry@gmail.com (Harry) (2010-06-08)
Re: Purple Dragon Book: Newbie questions on Chapter 4 text. Pidgeot18@verizon.invalid (Joshua Cranmer) (2010-06-09)
Re: Purple Dragon Book: Newbie questions on Chapter 4 text. felipeangriman@gmail.com (Felipe Angriman) (2010-06-09)
Re: Purple Dragon Book: Newbie questions on Chapter 4 text. simonsharry@gmail.com (Harry) (2010-06-10)
Re: Purple Dragon Book: Newbie questions on Chapter 4 text. cdodd@acm.org (Chris Dodd) (2010-06-12)
Re: Purple Dragon Book: Newbie questions on Chapter 4 text. felipeangriman@gmail.com (Felipe Angriman) (2010-06-13)
| List of all articles for this month |

From: Harry <simonsharry@gmail.com>
Newsgroups: comp.compilers
Date: Thu, 10 Jun 2010 21:21:30 -0700 (PDT)
Organization: Compilers Central
References: 10-06-023 10-06-025
Keywords: books, parse
Posted-Date: 13 Jun 2010 06:02:53 EDT

On Jun 10, 4:51 am, Joshua Cranmer <Pidgeo...@verizon.invalid> wrote:
> The reason that it's ambiguous is that if First(alpha) intersection
> Follow(A) is not empty, we can't tell if that terminal is part of alpha,
> and by consequent A, or if A is supposed to evaluate to epsilon and the
> terminal be part of Follow(A).




On Jun 10, 5:32 am, Felipe Angriman <felipeangri...@gmail.com> wrote:
>
> This means that when the parser need to make a choice between
> production rule 2 or 3 it will no be able to make such a choice
> because the select set are not disjoint.


Thanks, Josh and Felipe. FIRST(I) was having a difficulty but I think,
I FOLLOW(now). :-)


It seems, the three conditions will be necessary *only* when writing a
table-driven, predictive top-down parser and *not* when writing a
backtracking, top-down parser. Is this true, btw?



Post a followup to this message

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