ipp backend stops watching remote job too early

[michaelrsweet]

Version: 1.2.10
CUPS.org User: twaugh.redhat

The ipp backend monitors the progress of the remote job until:

job_state->values[0].integer > IPP_JOB_PROCESSING ||
job_state->values[0].integer == IPP_JOB_HELD

Unfortunately that means that applications trying to show the user what jobs they have (e.g. lpstat -o) don't show
a) remote IPP jobs currently held
b) remote IPP jobs currently stopped -- e.g. due to device error

..and they have no way to discover what happened to their job without knowing the network topology.

I think the condition should be changed to:

job_state->values[0].integer > IPP_JOB_STOPPED

What do you think?

[michaelrsweet]

CUPS.org User: twaugh.redhat

I actually would go further, and suggest that the ipp backend act as a proxy for the remote IPP job, stopping when it stops, restarting it when restarted, etc.

[michaelrsweet]

CUPS.org User: mike

It can't stop or restart on its own - the scheduler would need to do that, and that would require substantial changes to the code (without, IMHO, much benefit).

Stopping only after the job is completed (anything other than "stopped") sounds reasonable, however, so I will go ahead and make that change...

[michaelrsweet]

CUPS.org User: mike

Fixed in Subversion repository.

[michaelrsweet]

"str2352.patch":

Index: ipp.c

--- ipp.c (revision 6481)
+++ ipp.c (working copy)
@@ -1159,8 +1159,7 @@
* Stop polling if the job is finished or pending-held...
*/

•      if (job_state->values[0].integer > IPP_JOB_PROCESSING ||
  

•     job_state->values[0].integer == IPP_JOB_HELD)
  

•      if (job_state->values[0].integer > IPP_JOB_STOPPED)
  

  {
  if ((job_sheets = ippFindAttribute(response,
  "job-media-sheets-completed",