On Thu, May 09, 2019 at 05:11:35PM +0200, Pablo Neira Ayuso wrote:
> On Thu, May 09, 2019 at 05:11:06PM +0200, Pablo Neira Ayuso wrote:
> > On Thu, May 09, 2019 at 01:35:43PM +0200, Phil Sutter wrote:
> > > In scanner.l, that name is defined as alternative to "<=" symbol. To
> > > avoid the clash, it must be quoted on input.
> > >
> > > Fixes: 55715486efba4 ("proto: support for draft-ietf-tsvwg-le-phb-10.txt")
> > > Signed-off-by: Phil Sutter <phil@xxxxxx>
> > > ---
> > > Note that nft still produces invalid output since it doesn't quote
> > > symbol table values.
> >
> > I have reverted 55715486efba42 by now, I overlook that tests/py/ were
> > never run because the update for non-json is broken. @Logan: Please,
> > fix this and resubmit.
> >
> > BTW, a trick similar to what we do in primary_rhs_expr to deal with
> > the "le" token showing as a constant value will be needed.
I'm not sure if that's possible - I would expect shift-reduce conflicts
since the parser can't decide between 'ip6 dscp <= cs1' and 'ip6 dscp
le'.
> For the record, this 7/9 patch was left behind, not needed after the
> revert.
Sure, it wasn't a complete fix anyway.
Thanks, Phil
