[Fedora45] libcupsfilters‑2.2.1‑2 has a series of printing issues.

[szlt5]

Hi @zdohnal:

I downloaded Fedora‑Workstation‑Live‑45_Beta‑1.3 today and tested it, and found a series of printing issues.
Let me start with a brief overview. I will investigate the root causes in detail later, but Fedora 45 is about to be released, and having so many printing issues at this stage is quite frustrating.

[Test Driver]: Canon Linux UFRII printer driver (v6.40)

[Issue1]:
When printing on any printer that supports booklet mode, the output is still produced in booklet-layout even if single‑sided printing is selected.

[Issue2]:
The “Reverse Order” (outputorder=reverse) function has no effect.

[Issue3]:
A 4‑page document was printed with banner pages (start/end banner page are all set as "standard"), but only the banner pages and the first page were actually printed.

[Notes]:
These issues cannot be reproduced on Fedora 44. And cannot be reproduced when downgrading libcupsfilters from 2.2.1 to 2.1.1.

[zdohnal]

I'm sorry it looks like there are several regressions with switch from qpdf to pdfio. I did the basic testing whether the printing works by default, but I did not go into details.

[szlt5]

@zdohnal
I only performed a very rough test and already found these three issues. I’m not sure whether there will be additional printing problems later on.
[Issue 1] and [Issue 2] were caused by very minor reasons. I’ve already submitted a pull request #240, and Issue 3 is still under investigation.

[zdohnal]

@szlt5 new libcupsfilters is being built with your MR and scaling fix.

[ValdikSS]

Let me dump my LLM analysis here between the QPDF and PDFio code as well then. These are not necessary a bugs, just a difference.

1. Booklet comparison, already mentioned and already fixed
2. Booklet: missing shuffle-only mode support
3. Booklet: missing booklet-signature option support
4. number-up-layout support is missing
5. Scaling: missing cropfit, fit-to-page, ipp-attribute-fidelity option support
6. No pdftopdfAutoRotate / pdfAutoRotate options support, and ippo->pdf_auto_rotate never used anywhere, rotation is handled by scaling as it was previously PDFio pdftopdf does not fall back to autoprint print-scaling=auto #237, landscape-orientation-requested-preferred option missing
7. Output order: should also check page-delivery, and the additional value of reverse-order
8. Collate is on by default, sheet-collate option not handled
9. Hardware copies handling is absent & software copies are broken: hardware-copies / hardware-collate options not handled, even_duplex auto-enable for reverse+duplex is lost, %%PDFTOPDFNumCopies / %%PDFTOPDFCollate comments are never emitted [Regression 2.1.1 → 2.2.x] pdftopdf no longer emits its %%PDFTOPDF* marker comments, so libppd re-applies page-ranges and silently drops pages #233, pdftopdf (cfFilterPDFToPDF) no longer generates soft copies with *cupsManualCopies: True — regression in libcupsfilters 2.2.x #226
10. media-top/left/right/bottom-margin parsed only inside media-col (maybe the previous code was wrong?)
11. page-ranges: no leading - range support (QPDF handles -5 meaning 1–5)
12. Missing input-page-ranges support
13. Missing mirror-print alias of mirror option missing
14. Missing support for CLASSIFICATION environment variable (some kind of relic of the past?)

[tillkamppeter]

Let me dump my LLM analysis here between the QPDF and PDFio code as well then. These are not necessary a bugs, just a difference.

1. Booklet comparison, already mentioned and already fixed

[...]

14. Missing support for CLASSIFICATION environment [variable](https://github.com/OpenPrinting/libcupsfilters/blob/c726672e5bfd8b353f18a8c62ce27bc954552c16/cupsfilters/pdftopdf/pdftopdf.cxx#L409-L411) (some kind of relic of the past?)

I highly recommend to fix all these 14 points, to assure best backward compatibility.

@uddhavphatak Could you go through them?

[uddhavphatak]

Am looking into these and Will be providing a fix soon!

[ValdikSS]

I highly recommend to fix all these 14 points, to assure best backward compatibility.

I mentioned not all of these could be bugs in a sense that since the introduction of the older pdftopdf code, some options could be remapped to IPP options or to other options in different parts of CUPS and filtering now.

For example, cups ipp-mark and libppd ipp-mark remaps IPP collation options into CUPS Collate, and I guess it's not necessary for the filter to handle all of them now.

mirror is mapped to MirrorPrint, and the mention of mirror-print in the old pdftopdf looks like an IPP option, so maybe it should be remapped in cups+libppd instead?

Universal has also it's unique pipeline, I haven't checked that.

[szlt5]

[Issue3]:
A 4‑page document was printed with banner pages (start/end banner page are all set as "standard"), but only the banner pages and the first page were actually printed.

This issue was caused by my mistake. I entered the wrong input parameters. The banner page printing works fine.