[PAC] lack of PDF2 support

[davidcarlisle]

If presented with a PDF2/UA-2 file PAC shows a dialog warning that the PDF version is not supported

If you choose to carry on past this dialog it is validated to UA-1, eg the test document from #1302 produces a failed report:

As the non support is a documented feature that requires an active "OK" action from the user to check the PDF file despite being PDF 2 the isn't really a bug, however it does make it impossible to use PAC reliably for the default LaTeX PDF2 output.

If you drill in to the "more details" the specific failures are

• Lack of Alt text on Formula
• Lack of role mapping for "non standard" elements (reported for all MathML)

[teepeemm]

Related: PAC is aiming for PDF 2.0 and UA-2 in 2027: https://support.axes4.com/hc/en-us/community/posts/33625355546514/comments/33653836849298

[FrankMittelbach]

That's a long time in the future unfortunately; but perhaps more importantly, as seen in David's post: if one continues after the first warning then PAC gives a final verdict stating "This PDF is not PDF/UA compliant" while it should correctly say that "This PDF is not PDF/UA-1 compliant (but it identifies itself as PDF/UA-2 so it might be correct)".

Simply saying PDF/UA is totally misleading, but even saying "not compliant with PDF/UA-1" is misleading because people would still assume then that it is not valid for UA-2. Perhaps one can urge them to adjust at least their verdict statement before 2027.

[teepeemm]

In #1097, @FrankMittelbach asked us to "Please report PDF/UA-2 documents ...", but later said "or UA-1 if that is appropriate". Do you also want reports of valid UA-1 documents that produce spurious validation errors? At the moment, I'm recalling that PAC doesn't handle attribute classes correctly and so it ignores the Scope attribute for table headers.

[FrankMittelbach]

@teepeemm , yes, I think we should collect all issues that we identify with validation software, so if some software fails valid UA-1 then this should be listed here as well.