Re: [Extra] Francesca Palombini's No Objection on draft-ietf-extra-sieve-mailboxid-07: (with COMMENT)

[Alexey Melnikov <[e-mail]>]

On 16/03/2021 13:14, Ken Murchison wrote:

> On 3/16/21 8:50 AM, Bron Gondwana wrote:
>
>> Murray pointed out that I STILL had it wrong.  Uploaded again with a 
>> much better version :)
>>
>> MAILBOXID-OPT = ":mailboxid" objectid
>>
>> FCC-OPTS =/ MAILBOXID-OPT
>>
>> objectid is defined in RFC8474 as:
>>
>>        objectid = 1*255(ALPHA / DIGIT / "_" / "-")
>>                ; characters in object identifiers are case
>>                ; significant
>>
>> Of course now that I've posted this, I do wonder if I actually need 
>> to make it:
>>
>> MAILBOXID-OPT = ":mailboxid" SPACE DQUOTE objectid DQUOTE
>>
>> *sigh*.
>>
>> I mean, RFC8580 already doesn't seem to be very specific about the 
>> whitespace between tokens - but I expect we probably do want the 
>> mailboxid to be a string rather than an objectid for parser 
>> simplicity.  I welcome input on that!  (I'm pretty sure the parser we 
>> have in Cyrus right now needs it to be a string for example)
>
> I don't think adding SPACE is necessary, and no other extension does this.
>
Right, Sieve ABNF is applied to tokenized input, so whitespaces don't 
appear.
>
> We definitely want :mailboxid to take a string as argument, since 
> RFC5228 doesn't specify anything other than numbers, strings, and 
> stringlists.
>
> You can probably do something like what the imap4flags and special-use 
> extensions do and simply define:
>
> MAILBOXID-OPT = ":mailboxid" <mailboxid: string>
>
> and state in the text that the mailboxid string is limited to the 
> characters allowed in objectid.
>
Makes sense.
>
>   Of course, even if you don't state this, and the string contains 
> invalid objectid chars, the script will still parse as valid, but just 
> not match any mailbox.
>