Re: %destructor can leak the lookahead

[Tim Van Holder]

Paul Eggert wrote:
> Akim Demaille <address@hidden> writes:
>
> > - yydestruct can be called on EOF
> >  So what?  It should not be defined, and if the user meant to define
> >  it, there could be good reasons.
>
> Sorry, I don't follow.  Why should yydestruct not be defined here?

I think Akim means that the EOF token would probably not be "defined" in
the sense of having a semantic value, and it would not have a
%destructor entry.

> Sorry, I still don't follow.  Perhaps I'm misunderstanding the meaning
> of destructors in general.  From the documentation, destructors are
> called for all symbols that are "discarded by the parser".  But an
> ordinary reduce operation discards symbols, right?  So shouldn't
> destructors be invoked every time the stack shrinks, no matter why it
> shrinks?

During an ordinary reduce, the user has the opportunity to do something
with the symbols in an action - the same does not go for symbols the
parser discards internally.
Using yydestruct during reduce would be disastrous to many applications;
for example, many tools create AST-style nodes in the lexer and build
a tree from them in the parser.  If bison suddenly started destroying
the nodes they just put in the tree, that would be a Bad Thing(tm).
In fact, you could say those symbols were given to the user, not
'discarded' as such.