Re: Problems using bison and flex with C++

[Hans Aberg]

At 22:00 -0700 2004/06/26, Paul Eggert wrote:
>> It is the question whether each new invokation of the parser function
>> creates a new independent parser stack.
>
>Yes, it does.  (This is true even for non-pure parsers.)

Fine. It seems best to always make a new parser stack for each parser
invokation. I did not follow the development of the C parser since the time
it had a static array for a start, which then was global, and thus not
thread safe. Also, the C++ parser that Bison writes, and which I rewrote,
does not put the stack in the function, so it is not thread safe. If I
rewrite it sometime, I will change that.

  Hans Aberg