Re: [http-state] [Technical Errata Reported] RFC6265 (3444)

[Pete Resnick <[e-mail]>]

On 1/7/13 8:49 AM, Barry Leiba wrote:
>>>>> Corrected Text
>>>>> --------------
>>>>> path-value        = *<any CHAR except CTLs or ";">
>>>>> extension-av      = *<any CHAR except CTLs or ";">
>>>>>            
>>>> I would prefer if the quantifier is either pulled into the prose rule,
>>>> or the set of characters is put in a separate rule, say `av-octets`,
>>>> and the quantifier is then put in front of the reference to the rule.
>>>>          
>> Something like
>>
>>    path-value        = *av-octet
>>    extension-av      = *av-octet
>>    av-octet          =<any CHAR except CTLs or ";">
>>
>> Or
>>
>>    path-value        =<zero or more of: any CHAR except CTLs or ";">
>>    extension-av      =<zero or more of: any CHAR except CTLs or ";">
>>      
> The second is really a non-starter: we want to minimize the plaintext
> instructions and put as much as possible in actual ABNF syntax.
>
> The first seems fine, but why do you think it's better than what the
> errata suggests?  I don't seen any advantage (or disadvantage) either
> way.
>    

I tend to agree with Bjoern on this one. Though having a repeat before a 
prose-val is legal in ABNF, I find it pretty distasteful. Then again, I 
find prose-val in general to be distasteful. And I'm not clear on why 
you think the second suggestion above is a non-starter. Once we've gone 
and used prose-val, what harm is there in fixing the currently incorrect 
prose-val rather than adding a repeat to the front of it?

pr

-- 
Pete Resnick<http://www.qualcomm.com/~presnick/>
Qualcomm Technologies, Inc. - [phone]