Re: [dispatch] AD review of draft-holmberg-dispatch-iotl

[Christer Holmberg <[e-mail]>]

Hi,

>>>> """
>>>>   uri-parameter = transport-param / user-param / method-param / ttl-param
>>>>                   / maddr-param / lr-param / iotl-param / 
>>>> other-param """
>>>> You should check with someone more knowledgeable about ABNF than me, 
>>>> but I understand the better thing to do than redefining the 
>>>> uri-parameter production is to update it, e.g. "uri-parameter =/ iotl-param"
>>>
>>> Yes, that would be much clearer.
>>
>> AFAIK, that way of extending parameters have been deprecated, and one is (again) supposed to re-define the whole parameter.
>>
>> But, Paul can correct me if I'm wrong :)
>
> What made you think that? I know of no such change, *in general*.

Maybe I have misunderstood something, then - or I remember wrong.

> There has been some discussion about whether such a change (using =/) constitutes a revision to the base draft. (IMO it does, some 
> people don't think so.) If you actually do the change as you did then I think there is no question - it clearly is a revision!
>
> IMO, if you want to avoid such changes causing an update to the base document then you should do things such that the ABNF in the base document isn't modified.
>
> If there is a registry, then that really should serve as an alternative to listing values in ABNF. So then this issue doesn't come up.
>
> In this particular case, there now *is* a registry of URI parameters, and your draft *is* asking for it to be updated. So I was wrong in 
> suggesting that /= was better - I should have said that there should not be any change to the uri-parameter ABNF.

Ok. So, what shall I do in this particular case? :)

Regards,

Christer