[Printing-architecture] What should the "landscape" and "orientation-requested" options exactly do?

[Till Kamppeter]

On 12/07/2013 12:01 AM, Tobias Hoffmann wrote:
> I don't think so.
> E.g. for N-up, fitplot, adding of borders the pdftopdf filter looks at
> the final destination size. This size is determined by page-size,
> *together* with orientation-requested. pdftopdf also makes sure that
> filters like pdftoXYZ only get portrait pdfs as input, so that these can
> rasterize from to bottom and get the correct raster pixel ordering
> without thinking about rotations (e.g. an inkjet still prints along the
> short edge, regardless whether the input is landscape or not).
> 
> Two examples with texttopdf:
> 1)  A4, portrait: output pdf has (roughly) 21x29 cm, pdftopdf keeps it,
> pdftoijs renders (e.g.) 1000x1400 pixel output "image" for the printer.
> 2)  A4, landscape: output pdf has 29x21 cm, pdftopdf rotates it (in the
> ppd-specified normal_landscape direction), pdftoijs renders 1000x1400
> pixels.
> 
>   Tobias
> 

Thank you very much for your help. So now the handling of "landscape"
and "orientation-requested" in cups-filters should be correct: ...topdf
filters use it for generating layout, pdftopdf for n-up and fitplot. My
recent test files also show that pdftopdf does not break anything when
landscape" or "orientation-requested" is supplied and not n-up or fitplot.

Tobias, can you also have a look into fixing the handling of collate by
pdftopdf (see the other thread on this list)? Thanks.

   Till