Thanks for the response, Michael. We look forward to the discussion on September 11.
Soemin
-----Original Message-----
From: Michael Sweet <msweet at msweet.org>
Sent: Thursday, August 14, 2025 10:23 AM
To: PWG IPP Workgroup <ipp at pwg.org>
Cc: Soemin Tjong <stjong at exchange.microsoft.com>; Ibrahim Esmat <iesmat at microsoft.com>
Subject: [EXTERNAL] Re: [IPP] Request for Consideration: Updates to IPP INFRA Job Handling
[You don't often get email from msweet at msweet.org. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]
Soemin,
We'll need to find some time in one of the upcoming IPP workgroup conference calls to fully discuss this, perhaps the September 11, 2025 call since the agenda for the August 28th meeting is getting pretty full already...
Some initial comments from me below...
> On Aug 13, 2025, at 3:18 PM, Soemin Tjong via ipp <ipp at pwg.org> wrote:
> ...
> Challenge 1: Strict Ordering of Acknowledge-Job Before Fetch-Document
> Current Behavior:
> Section 5.3 of the IPP INFRA specification can be interpreted as requiring that Acknowledge-Job be called before Fetch-Document. If a job is already assigned to another Output Device, any subsequent Fetch-Document request results in a client-error-not-possible status.
I think we had made the ordering intentionally vague to allow for a Proxy to examine the Document(s) before committing to accepting the Job, so it would make sense to update the wording (at least) to make this explicitly clear.
> ...
> Challenge 2: No Mechanism to Release Acknowledged Jobs
>> Current Behavior:
> Once a job is acknowledged, it is irrevocably assigned to that Output Device, even if the device never prints it. This can lead to stalled jobs or confusion, especially in pull-print environments.
This is a bit trickier - I'm not keen on adding a new operation for this, but agree there should be a way to "recover" a Job so that another Proxy/Output Device can pick it up or finish it.
Certainly an implementation today could "time out" an acknowledgement if no Update-Job/Document-Status requests are received or revert the Acknowledge-Job/Document requests if the Proxy sends an Unregister-Output-Device request before marking the Job or Document(s) as completed, but there should be a way for a Proxy to explicitly return a Job to the fetchable state (maybe with Update-Job-Status?)
Think about the various error/exception cases and we can run through them during the conference call...
________________________
Michael Sweet