Re: set flex start conditions from bison?

[Akim Demaille]

Le 26 août 09 à 12:11, Rui Maciel a écrit :

> Akim Demaille wrote:
>
> > Can you show the content of this file?  Can you also show where you
> > include it in your Bison file?  Maybe you include the flex header  
> > from
> > the bison and vice versa, which probably does not work as expected.
> > Usually you include the Bison header from the Flex scanner (to get  
> > the
> > token definitions and so forth).  I have no experience with the
> > converse.
>
> I've sent an email with all the test project files attached (flex,  
> bison, makefile).

Please, keep everything public, send them here too.

I don't know what Flex you use, but using 2.5.35 it does not compile:  
you're using push_state, which is static in the scanner.c file, from  
the parser.c file.

Reading the Flex header file, I was most amused to read the following:

#define YY_FLEX_MAJOR_VERSION 2
#define YY_FLEX_MINOR_VERSION 5
#define YY_FLEX_SUBMINOR_VERSION 35
#if YY_FLEX_SUBMINOR_VERSION > 0
#define FLEX_BETA
#endif

Given their definitions of FLEX_BETA, I wonder if there ever existed a  
true release of Flex :)

Back to the point.  I do not have the problem you have: s_element is  
correctly replaced as expected.

address@hidden /tmp/foo $ grep '%[sx]' msh.l
%x s_element
address@hidden /tmp/foo $ grep s_element msh.y
        { yy_push_state(s_element); }
address@hidden /tmp/foo $ cc -E msh.tab.c | grep yy_push_state
    { yy_push_state(1); }


It seems bizarre to me to play with start conditions and yy_push_state  
from the parser.  I would rather write some wrapper function in the  
scanner.l file, say scan_enter_element_mode(), and use that function  
from my parser.

I agree it also seems bizarre that start conditions are defined in the  
header file, but that yy_push_state is static.