Re: Segmentation fault

[Tim Van Holder]

Arne 'Timwi' Heizmann wrote:
> Also, allow me to present a minimal test case which produces the
> segmentation fault:
>
> %token X
> %start startsymbol
> %%
> startsymbol       : /* empty */
> unusednonterminal : X
> %%

A segfault is definitely bad and needs to be fixed, but your grammar
is invalid - each rule needs to end with a semicolon.
I do seem to remember a segfault issue in the bison lexer being
patched - it may be this one.