On Fri, Feb 12, 2021 at 01:29:23PM +0100, Florian Westphal wrote: > Phil Sutter <phil@xxxxxx> wrote: > > > grammar bug. > > > > > > Pablo, Phil, others, can you remind me why we never did: > > > > Because this would be followed up by: > > > > | Subject: Unable to create a table called "trace" > > > > Jokes aside: > > > > I think Pablo didn't like the obvious consequence of having to quote > > *all* string types which are user-defined in output. He played with > > keeping the quotes as part of the name, so they are sent to kernel and > > in listing they would automatically appear quoted. I don't quite > > remember why this was problematic, though. > > > > In general, shells eating the quotes is problematic and users may not be > > aware of it. This includes scripts that mangle ruleset dumps by > > accident, etc. (Not sure if it is really a problem as we quote some > > strings already). > > Ok, but what if we just allow use of quotes in input? > That would at least allow to use nft to delete/add to chains created > by other tools. IIRC, this was deemed to make things worse as people may more easily create rulesets which break with 'nft list ruleset | nft -f -'. But that point won't hold anymore now, I guess. :D Extracting the changes to parser_bison.y from my patch in | Message-Id: <20190116184613.31698-1-phil@xxxxxx> might suffice already. Cheers, Phil
