Bill,
> On Jul 14, 2015, at 12:04 AM, wamwagner at comcast.net wrote:
> ...
> Michael,
> 1. I am unsure of the intent of your reference to RFC2119, and indeed find no use of the word “conditional” in that standard. To the contrary, item 1 states :
> “MUST This word, or the terms "REQUIRED" or "SHALL", mean that the definition is an absolute requirement of the specification. ”
Right, that was my point.
> 2. My problem with the 5100.6 definition is that it is unclear that the requirement is always for supporting a feature, and indeed it is unclear what the term “feature” encompasses.
I agree "feature" should be avoided. However, the 5100.6 definition also talks about the specified condition being true which I thought was the key change...
> So, unless there is objection, I suggest:
> CONDITIONALLY REQUIRED: A conformance requirement that applies if the specified condition is true.
I think we are in agreement. I'll prepare an updated template for review next Monday.
> With respect to Ira’s comment, I think that this is clear and concise. And if we are going to be consistent over PWG specs, the definition should not include the term “Printer”. Also, the rest of the compliance terms are defined by reference to RFC2119, which is general and deals with requirements.
>> But I do not have a strong position of this and will concur with the consensus.
> Thanks,
> Bill Wagner
>> From: Michael Sweet
> Sent: Monday, July 13, 2015 5:21 PM
> To: William A Wagner
> Cc: ipp
>> Bill,
>> Comments inline...
>> _________________________________________________________
> Michael Sweet, Senior Printing System Engineer, PWG Chair
>> On Jul 13, 2015, at 12:47 PM, wamwagner at comcast.net wrote:
> ...
>> I am unhappy with "... conformance requirement that applies to a particular capability or feature." as a general statement because:
>> a. Most conformance requirements apply to a particular capability or feature. I assume that one is to infer that the requirement applies only if that capability or feature exists in the subject implementation seeking to be conformant.
>> Per RFC 2119, REQUIRED, MUST, and MUST NOT are unconditional by themselves. You make them conditional by making the statement containing them conditional, e.g., 'If the Printer supports two-sided printing, the Printer MUST support the "sides" Job Template attribute.'
>> ...
>> There is there is the additional question as to whether the term should 'Conditionally Required" or "Conditionally Mandatory". Since the basic compliance term is "Required", "Conditionally Required" seems more appropriate.
>> Agreed.
>> A reasonable variation on the 2003 definition and the RFC2119 definition of requirement levels is :
>> CONDITIONALLY REQUIRED: definition is a requirement of the specification if the specified condition is true.
>> I'm not super happy with this...
>> One might prefer a term other than "definition", such as capability or feature, and consider a requirement of the implementation rather than of the specification, but RFC 2119 wording is:" … definition is an absolute requirement of the specification."
>> Alternatively, we could modify the current definition to:
>> CONDITIONALLY REQUIRED: A conformance requirement that applies if the specified condition is true.
>> I like this better, or the definition you quoted from 5100.6:
>> "The term CONDITIONALLY REQUIRED means that the Printer MUST support the feature, if the specified condition is true."
>>
_________________________________________________________
Michael Sweet, Senior Printing System Engineer, PWG Chair