Re: [Idr] [BULK] Bug in draft-ietf-idr-rfc5575bis, worth fixing?

[Jeffrey Haas <[e-mail]>]

Again, for the archive.

On Wed, Sep 30, 2020 at 08:51:17PM +0200, Robert Raszuk wrote:
> Hi Jakob,
> 
> > I personally prefer that BGP be nothing more than a transport for
> flowspec.
> 
> 100%
> 
> - - -
> 
> Can you take a look if XR, XE & NX all parse today's shipping FS NLRIs and
> validate each type and when applicable length to be "valid" ?
> 
> What all three of those implementations do when they:
> 
> a) run into the case of unknown type ?
> 
> b) run into the case of invalid value of any type carried inside the NLRI ?

A poll some time back when we were discussing the change to remove the
"opaque" behavior of several implementations suggested that no one dealt
with unknown components without a session reset.  This reduced the concerns
at the time that we were making a change that left implementations of -bis
as non-comformant.

This poll was certainly not comprehensive of all implementation, nor done in
such a way that the results should be posted officially.  An implementation
report might be appropriate.

-- Jeff