Re: [tcpm] draft-ietf-tcpm-rfc793bis-14: options

[Wesley Eddy <[e-mail]>]

With editor hat on, I fully agree with what Joe said.


On 7/31/2019 2:50 PM, Joe Touch wrote:
>
> Based on the defined scope and approach of 793bis, as stated in its intro:
>
> On 2019-07-30 23:26, [e-mail] wrote:
>
>> Hi Wes,
>>
>> Below some comments about the options as described in Section 3.1:
>>
>> * Consider adding some text to ACK the TCP option space problem. 
>> Adding a pointer to EDO as * an example * (without recommending it) 
>> of how to extend the space for non-SYNs would be useful.
> It would be useful to mention the ongoing work, but not to try to 
> repeat it.
>> * Add an explicit statement about 253/254 as being reserved for 
>> experiments with a strong recommendation to make use of shared ExIDs.
> It would be useful to note their being assigned for that purpose (not 
> quite RESERVED) and to ref 6994 for further details.
> Given the focus of this updated is standards-track, it would be useful 
> to remind developers not to deploy options using those codepoints 
> outside carefully controlled experiments and that they are default to 
> being not enabled in any code not under direct control of the developers.
> However...
>> * What about obsoleting (?) 6994 and incorporating its main 
>> contribution as part of 793bis?
> I disagree. First, 6994 is not being obsoleted per se; second, others 
> refer to its approach for other similar experimental fields. It would 
> be a bad idea to try to "obsolete" it.
> I also don't think it's useful to include these details inside 793bis 
> - these are not details needed for standards-compliant operation (esp. 
> because they shouldn't be used for final standards-track options).
> IMO, 6994 is a lot like congestion control; OK, if not better, as a 
> separate doc.
> Joe