Affected versions

• 91.0a1 (2021-06-02)
• 90.0b2
• 89.0

Affected platforms

• Windows 10x64
• macOS 11
• Ubuntu 18.04

Steps to reproduce

1. Launch Firefox.
2. Access https://bug1225439.bmoattachments.org/attachment.cgi?id=8688380 .
3. Open the print dialog.

Expected result

• Page is displayed correctly in the print preview window.

Actual result

• Page is displayed incorrectly in the print preview window.

Regression range

• Pushlog: https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=f0995e7b318af33c677fe0c0cae6616a6d8dfb3d&tochange=6754c6b70539e3cb7313cb5412f049a6e35a3aba
• Possible regressor: Bug 1702796

Additional notes

• see the attached screenshot.
• Actual print output is affected too (e.g. print to pdf)

Likely related to https://github.com/mozilla/pdf.js/pull/13120.
We should be probably not rotate the page but switch to landscape mode.
:Snuffleupagus, what do you think ?

This isn't really a regression as such, since this was changed intentionally to improve printing for (most) documents with landscape pages.

To fix this "properly", we need Firefox to support the @page CSS rule such that we can actually inform the browser about each page size/rotation when printing; hence any further improvements here are blocked on bug 851441 being fixed.

Can we perhaps disable this in Firefox until bug 851441 is fixed? Otherwise it doesn't seem like an improvement?

(In reply to Emilio Cobos Álvarez (:emilio) from comment #4)

Can we perhaps disable this in Firefox until bug 851441 is fixed? Otherwise it doesn't seem like an improvement?

For documents with mixed page orientations the current behaviour is a considerable improvement in most cases, hence I really don't see how re-introducing a very old and commonly reported bug would be a good idea here!?

Well, for only-landscape-orientation documents, which is more common, it is a regression, isn't it? E.g. the PDF in bug 1721127 prints upside down.

(In reply to Emilio Cobos Álvarez (:emilio) from comment #6)

Well, for only-landscape-orientation documents, which is more common, it is a regression, isn't it?

As far as I can tell it also depends on the Orientation setting in the print dialog itself, since if it's set to Portrait (which I guess is the default) then printing a PDF document with landscape pages should just work. Otherwise users would need to manually set the orientation to Landscape in the print dialog in order for the result to be correct (and the fact that this needed to be done manually was the source of many issues/bugs).

I suppose that it'd be possible to not rotate documents with only landscape pages, however it'd then re-introduce the old issue of forcing users to manually change the orientation in the print dialog. (Assuming of course that Portrait is the default setting.)

Given that the "correct" solution to all these issues seem to involve bug 851441 being implemented, any chance that it could be prioritized?

Bug 851441 is now fixed so we should be able to have a correct solution here.

The bug still exists with Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:100.0) Gecko/20100101 Firefox/100.0 Build:20220112163204

Let's mark this as a regression of bug 1702796, even if it was an intentional change that we want to keep (given that it fixed bug 844090, an S2, along with many other bugs), just for tracking purposes.

Set release status flags based on info from the regressing bug 1702796

Set release status flags based on info from the regressing bug 1702796

The PDF from comment 0 seems to be good now (probably fixed by bug 1820651), but this appears to still be reproducible in https://bug844090.bmoattachments.org/attachment.cgi?id=9298595.