Re: [nft PATCH v2] parser_bison: Accept arbitrary user-defined names by quoting

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Mon, Jul 01, 2019 at 08:13:41PM +0200, Pablo Neira Ayuso wrote:
> On Mon, Jul 01, 2019 at 06:11:39PM +0200, Phil Sutter wrote:
> > Hi Pablo,
> > 
> > On Fri, Jun 28, 2019 at 08:00:51PM +0200, Pablo Neira Ayuso wrote:
> > > On Mon, Jun 24, 2019 at 06:36:08PM +0200, Phil Sutter wrote:
> > > > Parser already allows to quote user-defined strings in some places to
> > > > avoid clashing with defined keywords, but not everywhere. Extend this
> > > > support further and add a test case for it.
> > > > 
> > > > Signed-off-by: Phil Sutter <phil@xxxxxx>
> > > > ---
> > > > Changes since v1:
> > > > - Fix testcase, I forgot to commit adjustments done to it.
> > > > 
> > > > Note: This is a reduced variant of "src: Quote user-defined names" sent
> > > >       back in January. Discussion was not conclusive regarding whether
> > > >       to quote these names on output or not, but I assume allowing for
> > > >       users to specify them by adding quotes is a step forward without
> > > >       drawbacks.
> > > 
> > > So this will fail later on, right?
> > > 
> > >         nft list ruleset > file.nft
> > >         nft -f file.nft
> > 
> > Yes, that's right. I sent a complete version which does the necessary
> > quoting on output in January[1], but discussion wasn't conclusive. You
> > had a different approach which accepts the quotes as part of the name
> > but you weren't happy with it, either. I *think* you wanted to search
> > for ways to solve this from within bison but we never got back to it
> > anymore.
> > 
> > This simplified patch is merely trying to make things consistent
> > regarding user-defined names. IIRC, I can already have an interface
> > named "month", use that in a netdev family chain declaration (quoted)
> > and 'nft list ruleset' will print it unquoted, so it can't be applied
> > anymore. Without my patch, it is simply impossible to use certain
> > recognized keywords as names for tables, chains, etc., even if one
> > accepted the implications it has.
> 
> I'm not arguing there's something to fix.
> 
> I'm telling this is still incomplete.
> 
> Would you allocate a bit of time to discuss this during the NFWS?

I mean, a quick summary of the different options for a complete
solution for this, and we decide there.

Unless you tell me this is very urgent :-)



[Index of Archives]     [Netfitler Users]     [Berkeley Packet Filter]     [LARTC]     [Bugtraq]     [Yosemite Forum]

  Powered by Linux