Webinar October 2022 📍 Online

PDF Association TechConnect 2022 — Virtual: PDF 2.0 New Features for Implementors

The PDF Association TechConnect 2022 was a virtual event aimed at PDF library and tool implementors. It provided a systematic walkthrough of PDF 2.0 new features that differed from PDF 1.7, focusing on the changes most relevant to library developers: new object types, encryption updates, associated files, and structure element additions.

Official event page →

Key takeaways

  • AES-256 GCM encryption (AEAD, authenticated) and unencrypted wrapper documents were the major new PDF 2.0 encryption additions — both had significant implementation gaps in the 2022 testing results.
  • XFA forms deprecated in PDF 2.0 — new tool development should target AcroForms; existing XFA processing must be maintained for legacy compatibility.
  • Associated files (/AF) on objects other than the document catalogue was the most consistent cross-vendor implementation gap found in the 2022 interoperability testing.
  • Cross-reference streams, object streams, and basic tagged PDF were all consistently implemented — these features had had many years to mature.
The PDF Association TechConnect 2022 was a virtual developer event held in October 2022, organised by the PDF Association. Following PDF Days Europe 2022's return to in-person and its introduction of the PDF/UA-2 draft, TechConnect 2022 focused on the implementation-level details of PDF 2.0 features that library and tool developers needed to support. PDF 2.0 (ISO 32000-2:2017) had been published five years before TechConnect 2022. While most major PDF libraries had added basic PDF 2.0 support, the depth of implementation varied — particularly for the new features that differed most significantly from PDF 1.7. TechConnect 2022 provided a systematic walkthrough of the PDF 2.0 changes most relevant to developers, with practical implementation guidance derived from the PDF Association's interoperability testing programme.

PDF 2.0 new features: a developer walkthrough

New encryption algorithms — PDF 2.0 added AES-256 in GCM (Galois/Counter Mode) as a permitted encryption algorithm for PDF encryption, alongside the existing AES-256 in CBC mode. GCM provides authenticated encryption (AEAD) — the cipher text includes an authentication tag that detects tampering. Implementation sessions covered the specific PDF encryption dictionary changes for GCM: the /CF (Crypt Filter) dictionary now includes /AuthEvent and a method value of AESV3 with a GCM flag. Unencrypted wrapper documents — PDF 2.0 introduced the concept of an unencrypted wrapper document that contains an encrypted payload PDF. The wrapper is a valid unencrypted PDF with a cover page explaining that the content is encrypted. The /EncryptedPayload file specification dictionary in the wrapper references the encrypted PDF. This allows the wrapper to be indexed and its metadata to be read without decryption, while protecting the actual content. Sessions reviewed the implementation requirements and edge cases. Page-level metadata streams — PDF 2.0 allows each page to have its own XMP metadata stream (/Metadata on the page dictionary), enabling per-page metadata for documents that aggregate pages with different copyright, provenance, or language metadata. Sessions covered the expected validator and viewer behaviour when page-level metadata is present. Associated files at the object level — PDF 2.0 normalised the PDF/A-3 associated file mechanism: any PDF object can now have an /AF (Associated Files) array. Sessions covered the handling requirements: parsers that only check the document catalogue for /AF miss associated files on pages, annotations, and structure elements. Document part metadata (/DPart) — The /DPart hierarchy was introduced for variable data printing (VDP) workflows where a single PDF represents multiple personalised document instances (mailings, tickets, invoices). Each document part has its own /DPartRoot tree entry with per-instance metadata. Sessions covered the tree structure and the interaction with /AF for per-instance embedded files.

Deprecated and removed features in PDF 2.0

PDF 2.0 deprecated or removed several features from PDF 1.7. Implementations needed to handle both producing documents that avoided deprecated features and consuming existing documents that still used them. Deprecated encryption — RC4 encryption (PDF 1.x standard encryption with 40-bit and 128-bit keys) was deprecated in PDF 2.0. PDF 2.0 generators must not produce RC4-encrypted PDFs, but parsers must continue to support them for compatibility with existing files. Deprecated JavaScript structure — Some JavaScript attachment mechanisms from earlier PDF versions were deprecated in favour of the PDF 2.0 JavaScript name tree structure. Sessions covered the migration path for implementations that need to produce PDF 2.0 compatible JavaScript structures. XFA forms deprecated — PDF 2.0 deprecated XFA (XML Forms Architecture) forms, which had been Adobe's proprietary dynamic form format. XFA required the Adobe LiveCycle engine for full functionality and was never part of the ISO 32000 standard (it was a non-normative reference). Deprecation means: new PDF 2.0 generators should not produce XFA forms (use AcroForms instead); parsers must continue to parse XFA forms in existing documents. Sessions covered the migration path for XFA forms to AcroForms for developers working with legacy form data. Type 1 fonts in PDF 2.0 — Adobe Type 1 fonts (the PostScript-era font format) were deprecated in PDF 2.0 in favour of OpenType (CFF-based or TrueType-based). Sessions covered font subsetting and embedding requirements for OpenType fonts, which differ from Type 1 subsetting in several edge cases.

Interoperability testing: lessons for implementors

TechConnect 2022 presented preliminary results from the PDF Association's PDF 2.0 interoperability testing programme — the first systematic cross-vendor testing of PDF 2.0 implementations. Testing methodology — The PDF Association assembled a corpus of test PDF files specifically designed to exercise PDF 2.0 features: each feature was covered by multiple test files that varied parameters, exercised edge cases, and included both valid and intentionally invalid examples (to test parser error handling). Consistent results — Object streams, cross-reference streams, and basic tagged PDF structure were handled consistently across all participating implementations. These features had been in PDF since PDF 1.5 (2003) and had had many years of implementation experience. Divergence areas: - AES-GCM encryption: Implementation quality varied significantly; some implementations had subtle GCM authentication tag handling errors. - Unencrypted wrapper documents: Several implementations did not implement this feature at all; others implemented it but with incompatible interpretations of the /EncryptedPayload dictionary structure. - Page-level /Metadata: Inconsistent handling — some parsers ignored page-level metadata; others correctly prioritised it over document-level metadata. - Associated files on objects other than the document catalogue: The most consistent divergence point, confirming a systematic implementation gap. Action items — Implementors received specific test files for each divergence category and committed to addressing the identified issues in future releases. The PDF Association's testing infrastructure allowed ongoing regression testing to track progress.

Frequently asked questions

What is AES-GCM and why is it used in PDF 2.0 encryption?

AES-GCM (AES in Galois/Counter Mode) is an authenticated encryption mode that provides both confidentiality (like AES-CBC) and integrity/authenticity (unlike AES-CBC). In AES-GCM, the encryption process produces not only cipher text but also an authentication tag — a short value that any would-be tamperer cannot forge without knowing the key. When decrypting, the recipient verifies the authentication tag before trusting the plaintext. In AES-CBC (the older PDF 1.7 encryption mode), an attacker who modifies cipher text bytes can cause decryption to produce modified plaintext without detection — a property exploited in padding oracle attacks. AES-GCM eliminates this class of attack. PDF 2.0's addition of AES-GCM encryption is primarily relevant for high-security document workflows where tamper detection alongside confidentiality is required.

What is XFA and why was it deprecated?

XFA (XML Forms Architecture) was Adobe's proprietary format for dynamic PDF forms, introduced in LiveCycle Designer. An XFA form embeds XML data streams within a PDF that describe form layout, data binding, and JavaScript-based behaviour. XFA enabled sophisticated dynamic forms that could reflow, grow, and interact with backend systems — capabilities beyond standard AcroForms. However, XFA required Adobe's LiveCycle runtime for full functionality — standard PDF viewers showed only a static rendering. XFA was never part of the ISO 32000 standard (it was referenced normatively in PDF 1.7 but specified in separate Adobe documents). PDF 2.0 deprecated XFA because: it was not interoperable across non-Adobe viewers; it required proprietary Adobe infrastructure; and modern HTML5 forms, properly designed AcroForms, and accessible web-based forms could serve the same purposes with broader compatibility.

What is an unencrypted wrapper document in PDF 2.0?

An unencrypted wrapper document is a PDF 2.0 feature that allows an encrypted PDF to be wrapped inside an unencrypted container PDF. The wrapper PDF is a fully valid, unencrypted PDF — typically with a cover page explaining that the content is protected. It contains an /EncryptedPayload file specification that references the actual encrypted PDF, which is embedded as a file attachment. The practical benefit: document management systems, search engines, and email gateways can index the wrapper's metadata and cover text (author, title, keywords, document description) without having the decryption key. The encrypted payload is protected, but its existence and basic metadata are discoverable. This architecture is useful for rights-managed documents, secure delivery of sensitive documents to specific recipients, and DRM workflows where document discovery and content protection need to be separated.

Try it in your browser

The PDF tools and standards discussed at this event are available free — no account needed.