ietf-nntp newsgroup production

[Chin Chee-Kai]

Thanks for including USEFOR definitions, Charles.


>> In <Pine.LNX.3.93.1000726124639.1130C-100000 at one.softml.net> Chin Chee-Kai <cheekai at SoftML.net> writes:
>> >+         The conventions as defined in RFC822 are used.  In addition, the
>> >+         notation "\", when used in the production rules, denotes exclusion.
>> 
>> No, I son't think you want to invent a special notation just so you can
>> use it in one situation.

Well, I can see why you say that.  It's a convenience suggestion.
Alternatively, we could list the range of characters covered in
"parameter \ DOT" and name it as a new character set for use in
newsgroup-segment.



>> >+           argument = parameter ; excluding sequence ".."
>> >+           ;;; original production ::  augument = parameter ; excluding sequence ".."
>> 
>> But since 'argument' is not used elsewhere in the syntax, why not leave it
>> out?

Because "argument" is used elsewhere in the descriptions in the
earlier sections.  The intro paragraph of Section 13 says so.
Someone tell me my interpretation is wrong.



>> >+           newsgroup = newsgroup-segment *(DOT newsgroup-segment)
>> >+           newsgroup-segment = parameter \ DOT
>> >+           DOT = %x2E
>>
>> The USEFOR syntax is:
>> 
>> newsgroup-name  = component *( "." component )
>> component       = component-start
>>              *( component-start / component-other )
>> component-start = Un-lowercase / Un-digit
>> Un-lowercase    = <Unicode Letter, Lowercase> /
>>           <Unicode Letter, Other>
>> Un-digit        = <Unicode Number, Decimal Digit> /
>>           <Unicode Number, Other>
>> component-other = "+" / "-" / "_"
>>
>> The complication arises because of the UTF-8 and a desire to exclude
>> upparcase letters. I do not think we need quite so much complication in
>> this document, so I would suggest:
>>
>> newsgroup	= component *( "." component )
>> component 	= <a sequence composed of letters, digits,
>> 		  "+", "-" and "_">
>> 

Ah, the USEFOR syntax doesn't quite work.  The "component-start" 
production does not allow "-" to be used as a non-starting character 
in the first component of a newsgroup reference.  Currently, just off
my neighborhood ISP's news server, I already see 
          "relcom-list.internic.net-happening", and 
          "z-netz.*"
as counter-examples.  So Charles, your suggestion is fine, except 
you gotta add the necessary definitions for "letters" and "digits"
(though obvious), and also the 0x80-0xFF characters for UTF-8
multi-lingual newsgroup references.


Cheers,
CK