On Tue, Aug 5, 2025 at 7:52 AM Jen Linkova <[email protected]> wrote:
> Hi Bill, > > On Tue, Aug 5, 2025 at 8:19 AM Bill Fenner <[email protected]> wrote: > >> A question: if [I-D.ietf-intarea-rfc8335bis] acknowledges that > inferring the > >> length is suboptimal, and this draft solves the issue: what prevents us > from > >> using the mechanism defined in THIS document for > [I-D.ietf-intarea-rfc8335bis]? > >> Is it just because this draft was written before > I-D.ietf-intarea-rfc8335bis? > >> Now this draft seems to "overtake" [I-D.ietf-intarea-rfc8335bis], so I'm > >> wondering: if it gest published before [I-D.ietf-intarea-rfc8335bis] - > would it > >> make sense for [I-D.ietf-intarea-rfc8335bis] to use it? > > > > > > When I started the rfc8335bis update, I started with an implementation > survey. None of the implementations that I could find actually implemented > what RFC8335 + RFC4884 specify : in particular, they all added extra data > after the extension objects, where RFC4884 implies that the extension > objects extend to the end of the packet. > > > > We decided that my update would reflect with the deployed > implementations, as opposed to trying to say that they're incorrect (by > strictly requiring RFC4884 behavior) or incomplete (by defining an RFC4884 > object for "extra data after the ping"). > > > > Defining RFC8335bis to use icmp-exten-hdr-len is contrary to that > decision. If we have carte blanche to redefine the PROBE packet format, I'd > rather just have defined an RFC4884 object for extra data and use that. > > Ah yes, thanks for refreshing my memory on this, I did forget the > details of that story. > It all makes sense but there is no reason for future implementations > of the PROBE to also use the length field, right? > (Let's say I'm writing a new library to support ICMP extensions, after > draft-ietf-intarea-icmp-exten-hdr-len is published. Now, when crafting > the ICMP extension header for a packet, I'd either need to handle > PROBE packets differently from any other packets (set length = 0) or > just set the length). > Anyway this discussion is more in scope for rfc8335bis than for > draft-ietf-intarea-icmp-exten-hdr-len. > Ok. RFC8335bis just says "use RFC4884" and says almost nothing about the 4884 packet format, so as long as draft-ietf-intarea-icmp-exten-hdr-len updates RFC4884, I don't know that there's anything to say in 8335bis. Bill
_______________________________________________ Int-area mailing list -- [email protected] To unsubscribe send an email to [email protected]
