Update on the Resolution of Ambiguities

In December of last year we reported the development of the PDF Validation TWG’s Resolution of Ambiguities document, with an additional 10 questions added to the 4 previously presented to the ISO committee and resolved in April, 2015 during the meetings in San Jose, California.

Since last November the veraPDF contractor has raised, and the TWG has addressed, several more ambiguities to the PDF Validation TWG for resolution, bringing the total number of ambiguities raised to 24 for all parts of PDF/A.

Since many of these questions pertain to PDF/A-next in addition to previous Parts of ISO 19005, the 10 new questions generated by the TWG between the two ISO meetings were submitted into the formal ISO process for reviewing comments against draft specifications. The ISO WG then duly considered the Resolution of Ambiguities document during its meetings in Ghent, Belgium in May, 2016.

These new questions proved somewhat more contentious than many of the questions formerly raised. To provide a flavor of the issues addressed, the most recent set of ambiguities is summarized here:

veraPDF-A015 discusses the interpretation of the corrigendum 2 to ISO 19005-1, which contains a special clause to exclude resources unreferenced from the corresponding content stream from further requirements.

At Adobe’s request, this item was parked by the WG for further study, to be resolved at the ISO meetings in Sydney, in November, 2016.

veraPDF-A016 remains a sore-spot. The keys in question are deprecated from ISO 32000-2, and thus do not affect PDF/A-next. However, the requirement remains for PDF/A-2 and PDF/A-3; it will be left to an industry Application Note to provide a universal reference for relaxing these unnecessary and problematic requirements for CharSet and CIDSet entries.

veraPDF-A017 sought to clarify that XMP metadata streams in PDF/A-1 must be uncompressed. The TWG’s interpretation was accepted, and the WG added an additional clarification: that XMP packages don’t need to conform to XMP or even XML.

veraPDF-A018 refers to an ambiguity over whether the requirement pertains to the file-format or to a means of comparing real values. The WG decided that Non-zero values less than the minimal one are not allowed in PDF/A-2 (and PDFA-3) on purpose.

veraPDF-A019 discusses the problem that clause 6.1.13 in ISO 19005-2 copies the list of limits from ISO 32000-1 and lists them explicitly. However, the word “approximately” was dropped, and so the definition of the limits thus differs between ISO 32000-1 and PDF/A-2, creating an untenable situation for processors encountering files that (may) exceed these limits. The WG elected to leave the matter as-is because although differing from the base specification for PDF the actual requirement for PDF/A-2 was itself not ambiguous.

veraPDF-A020 concerns the “shall” requirement in all three parts of PDF/A to comply with either predefined schemas from the XMP specifications or with an extension schema. The WG accepted the PDF Validation TWG’s recommendation for PD/A-next.

veraPDF-A021 questions the value and practicality of the requirement in PDF/A-2 and PDF/A-3 to record user actions in the xmpMM:History property. The WG accepted the PDF Validation TWG’s recommendation for PD/A-next but highlighted that the parameters field is still required in xmpMM:History for conformance with PDF/A-2 and PDF/A-3.

veraPDF-A021a (there was a numbering error, to be corrected in a subsequent Resolutions document) points out that in PDF/A-1 it’s not clear if any Widget annotation is required to have an annotation dictionary. The WG agreed with the TWG’s interpretation that for PDF/A-1, every button field widget shall have an appearance stream or dictionary.

veraPDF-A022 affects all parts of PDF/A. The requirement for multiple appearance streams misses the case when a form (such as a radio button) has multiple widgets associated to it and defined in /Kids array. The TWG proposed to PASS otherwise valid PDF/A documents if it contains a Widget annotation dictionary with Parent key referring to a parent form field of type Button, and if the value of the N key in this widget annotation dictionary refers to an appearance subdictionary. The WG agreed.

veraPDF-A023 pointed out that some wording pertaining to ICC color spaces was imprecise, and proposed specific replacement text. The WG accepted this interpretation, and the PDF/A-next Project Leader agreed to make this change in the text of PDF/A-next.

Following the ISO meetings in Ghent the PDF Validation TWG will continue its review and test-suite development for PDF/A-2 and PDF/A-3, with its final questions to be put before the ISO WG during the November, 2016 meetings in Sydney, Australia.

The PDF Association is currently considering publication of the final Resolution of Ambiguities document as a formal PDF Association Application Note for PDF/A.