I think you are right, that the 'unknown' job state should be used only when
the agent doesn't know about the state of the job that exists. When the job
doesn't exist anymore (or before the job exists) is not really a job state at
all. I think that this problem can be easily fixed by changing the heading on
the column from 'unknown' to something like <job doesn't exist>, ok?
Tom
At 14:02 05/21/97 PDT, Bill Wagner wrote:
>I apologize if I missed a step or two in this exchange. Of course, I agree
>with the objective and believe the states listed are adequate. However, the
>'unknown' state, within general MIB use at least, has been more a
>reflection on the state of the equipment than the job. That is, the state
>of a job is unknown because of some sensing difficulty or sequence
>difficulty with the equipment, not because it is a normal state in the
>sequence of handling a job.
>
>It would be quite possible for a unit to report no state other than
>unknown. It seems it would also be possible for a unit to recognize a state
>after it has previously been unknown. For example, if a job can be
>identified, it can be canceled and the state is then likely to be canceled
>and not unknown. In short, I suggest that unknown is a wild card rather
>than a job state that exists before pending and processing and after
>canceled and completed.
>
>If something is necessary for the pre-pending and post complete states,
>what about Other?
>
>Bill Wagner, Osicom/DPI
>
>
>______________________________ Reply Separator
_________________________________
>Subject: IPP> MOD - simplified legal state transition diagram
>Author: Tom Hastings <hastings@cp10.es.xerox.com> at Internet
>Date: 5/21/97 7:29 AM
>
>
>So the simplified legal state transition diagram for IPP and JMP would become,
>following Bob's proposal:
>
> New state
>Old state unknown(2) pending(3) processing(4) canceled(5) completed(6)
>unknown(2) yes yes
>pending(3) yes yes
>processing(4) yes yes yes
>canceled(5) yes
>completed(6) yes
>
>A blank entry indicates impossible transitions. Self loops are not
>indicated, such as a Get-Attributes operation on a job, since they aren't
>a job state transition.
>
>For IPP we need to consider which job states and job state transtions are
>required for conformance and which are optional.
>
>JMP requires that processing(4), canceled(5), and completed(6) are
>mandatory, so that pending(3) is optional. What about unknown(2)?
>The former needsAttention state was also mandatory, so that the
>new job-state reason: printer-stopped, should become a mandator job state
>reason, correct?
>
>For IPP and JMP, all transitions into the canceled state should be required.
>However, the transition from processing back to pending (a job-held situation)
>is optional.
>
>
>