> On 11 Oct 2016, at 10:24, guillaume marques <[email protected]>
> wrote:
> I have a little question regarding an example in the Bison's documentation.
>
> In the Section 7.2 Lexical Tie-ins, how come the provided example does what
> it is explained to do?
>
> I understand the principle of the flags...
Bison implements an extra implicit rule:
expr: HEX ’(’ X1 expr ’)’
...
;
X1:
/* empty */ { hexflag = 1; }
;
Because this rule is empty, and the parser may or may not make a lookahead, the
result can be unexpected.
> ...but I don't understand how Bison can
> understand when to parse the integer as an hexadecimal over an identifier
> since we never say, when the flag is nonzero, that we should parse an
> hexadecimal,
> and when it is zero, to parse an identifier.
>
> I only see that the flags are set but not used.
It is handled by the lexer (cf. the last sentence of that paragraph):
In appropriate places, the variable is read, and different lexing takes place.
Flex admits start conditions that can be set before the lexing begins.
_______________________________________________
[email protected] https://lists.gnu.org/mailman/listinfo/help-bison
