Re: token extraction problem within a rule

[Hans Aberg]

On 8 Nov 2007, at 10:49, Richard Gill wrote:

> I started learning flex/bison some days ago to write a parser. Nearly
> all my parser works well, but I've a problem with a rule which matches
> well but gives a wrong result in $n. I checked the lexer (flex) did
> extract right token.

There is a quite common newbie error to not copy the token that the  
Flex generated lexer points. It just points to a buffer, temporarily  
'\0' terminated, and changed back. Bison then makes several calls,  
and if given just a pointer in the buffer, there will be some jumble.

  Hans Aberg