[cups.general] Weird behaviour of some HP printers

[Helge Blischke]

Michael Sweet wrote:
> Helge Blischke wrote:
> 
>> Michael R Sweet wrote:
>>
>>> Helge Blischke wrote:
>>>
>>>> When upgrading from CUPS 1.1.19 to CUPS 1.3.5, we encountered quite a
>>>> weird thing with some HP printers, especially the HP 4250 series:
>>>>
>>>> When configuring the printer to use the "Fit to Page" option
>>>> to "Nearest Size and Scale", it worked as expected in 1.1.19
>>>> but failed to scale down the pages in 1.3.5.
>>>>
>>>> A detailed analysis revealed that the only significant difference is 
>>>> the
>>>> order of the setpagedevice statements in the output of the
>>>> pstops filter (short of line endings and comment lines).
>>>>
>>>> Is there anybody out there who has a reasonable explanation?
>>>
>>>
>>>
>>> No, very weird.
>>>
>>> Can you provide diffs (diff -u) between the two versions?
>>>
>>
>> See the attachment.
>>
>> Just to forestall a question I suspect:
>> the CUPSsetpagedevice stuff in the 1.3.5 variant redefines
>> the setpagedevice operator to merely collect all the setpagedevice
>> settings into one single dictionary and execute the real
>> setpagedevice operator with this dictionary just at the
>> end of the setop section of the job.
> 
>  > ...
> 
> We don't do this, as it will interfere with proper ordering of
> printer feature commands - are you sure you are using the standard
> pstops filter?
> 

Yes and no, but when the collecion stuff is disabled, the order of
the statements is exactly the same as with the original filter.

BTW, according to the PLRM, 3rd edition, the order of setpagedevice
invocations is irrelevant as long as no marks have been put onto the
page. We do the collecting to avoid watermark flickering on the first
page.

Helge

NB: All other printers but the HP 4250's (we have three of them here)
   work correctly.

-- 
Helge Blischke
Softwareentwicklung

H.Blischke at acm.org