[cups.general] back to square one with job killing

[Ambrose Li]

On 21/12/06, Brandon S. Allbery KF8NH <allbery at ece.cmu.edu> wrote:
>
> On Dec 21, 2006, at 18:16 , Gene Heskett wrote:
>
> > I suspect that may have been the case here, but this is a mere
> > Epson C82,
> > whose memory buffer probably can't hold more that one sweep of the
> > head
> > out & back.  It actually surprised me that it could print ascii text!
>
> The other possibility is that the OS itself is doing massive
> buffering, in which case there *may* be some system-dependent way to
> make it flush the buffer; but I never found one on the few unixlikes
> I used which had that kind of antisocial behavior, aside from
> rebooting.  It's hard for me to fault CUPS in that case.

At least for CUPS 1.x, I find that a more reliable way to completely kill
a job is:

- power off printer
- shut down cups
- do a "ps", "lsof", etc. to look for any backends that fail to shut down,
   then manually kill them
- make sure the backends really are killed, otherwise repeat previous step
- power on printer
- restart cups

As for who to fault, failing to kill the backends when CUPS itself shuts
down is, IMHO, at least in part, CUPS' fault.

-- 
cheers,
-ambrose