Re: [Foomatic] Re: [Gimp-print-devel] Re: [lsm-printing] Printer drivers: Clean Cancellation urgently needed!

[Enrique Perez-T. <en...@on...>]

On Wed, 27 Jul 2005 12:45:37 +0200, Johannes Meixner <js...@su...>  
wrote:

> On Jul 23 13:37 Michael R Sweet wrote (shortened):
>> Johannes Meixner wrote:
>> > - For "raw" queues there is no driver.
>>
>> I don't think that represents the typical CUPS user that is
>> asking for better cancellation support.
>
> When printing from Windows client systems to a CUPS server
> the printer specific data is often made by the Windows driver
> and then the CUPS server does only "raw" printing.
>
> I agree that "clean cancellation" even from Windows (or whatever
> other operating systems) may lead to endless problems.

This seems to confirm that cancelleation cannot be seen purely as
a property of the driver rendering the current job.

The architecture should have means of informing the renderer where
this is feasible, but ultimatly the spooler that "owns" the printer
(where the printer is locally connected) should have a last word.

In cases where there is no obvious owner (network printers) there
are two possibilities.  Network printers are really dedicated
computers built into the printer case, and all depends on the
protocol offered by the software provided in that computer.

Either printing on them should be handled as a remote print job
where the cancellation request is forwarded to the build-in server
as the protocol permits,  or the printer is handled as a local one
in spite of the connecton technology, and the local spooler, or an
designated one, must initiate a cleaning job.

Cleanup jobs are of course printer-specific, with the ppd containing
the necessary information to select the right cleaning driver.

I think that for many printers, closing the connection and reopening
it brings the printer's protocol engine to a known state, but the
printer as a whole may be in an error state with a sheet of paper
half through. Then a cleanup job will be one that ejects that page.

(In any case, it may be usefull to distinguish two kinds of "known
states" for the printer, one relative to the communication protocol
or language, the other w.r.t. the printing engine itself.)

Regards,
Enrique