Re: Useless nonterminals - what and why?

[Hans Aberg]

At 11:26 +0100 0-11-20, Hans Aberg wrote:
>Think of the rule as written (as when debugging is turned on)
>  http_proxy_list <- http_proxy_list http_proxy
>withe the ":" replaced by an arrow "<". Then in order to find a LHS (left
>hand side) http_proxy_list, you need to first find a RHS http_proxy_list
>followed by a http_proxy.

I have noticed that some books on compilers write the rules with the arrow
the other direction
(*)  http_proxy_list -> http_proxy_list http_proxy
I think this is because one can think of a language as all sentences that
can be generated from LHS side rules into RHS side rules. -- By contrast,
Bison has a given sentence and tries to reduce it to the sentence symbol.

Then the reasoning becomes: In order to apply (*), one must first generate
a http_proxy_list. But there is no other rule by which a http_proxy_list
can be generated; so the rule is useless.

  Hans Aberg