Re: Reductions during Bison error handling

[Akim Demaille]

>>>>> "Paul" == Paul Hilfinger <address@hidden> writes:

Paul> It passes the regression tests and seems equivalent to mine.  I
Paul> have no objection, although that has to be one of the more
Paul> impressive 'while' statements I've seen in some time!

Same here.

I like the patch too.  I'm also happy to see the skeleton being
simplified, as it will make it easier later to have a `push' version
of it, instead of providing only the `pull model' (calling yylex).

Please, whoever wants, install this patch!  I would ask to the
installer to report this change in NEWS.  And if possible to adjust
the C++ parser too (otherwise we'll do it, so don't feel obliged to).