Conference October 2021 📍 Online

PDF Association TechConnect 2021 — Virtual: PDF 2.0 Corrigendum and Pandemic-Era Workflows

PDF Association TechConnect 2021 was the first dedicated developer-focused virtual event, held in October 2021. Sessions covered the ISO 32000-2 first corrigendum (clarifying ambiguities discovered in early PDF 2.0 implementations), the surge in cloud-based PDF processing driven by remote work, and early PDF/UA-2 technical drafts.

Official event page →

Key takeaways

TechConnect 2021 was the first PDF Association developer-specific event — signalling growth of the developer community
ISO 32000-2 first corrigendum addressed /AF, transparency, optional content, and page-level output intent ambiguities
COVID-19 pandemic drove unprecedented cloud PDF processing adoption; e-signing volumes more than doubled in 2020
Chromium headless print-to-PDF identified as not suitable for accessibility/archive workflows without remediation
PDF/UA-2 committee draft available — /B deprecation, MathML via /AF, and namespace mechanism previewed
Developer feedback on the PDF/UA-2 draft influenced the final specification

PDF Association TechConnect 2021 was the first dedicated developer-focused virtual event from the PDF Association, held in October 2021. Unlike the annual PDF Days events — which blend developer, vendor, and end-user audiences — TechConnect was targeted specifically at software engineers and technical architects building PDF processing tools, document management systems, and standards-compliant applications. The 2021 edition arrived in a period of compressed change: the COVID-19 pandemic had accelerated remote work and cloud document processing at a pace that would otherwise have taken years; PDF 2.0 had been published in 2017 and corrigendum corrections were now available; and PDF/UA-2 was in active development with an early draft available for community review.

ISO 32000-2 first corrigendum: clarifying PDF 2.0

The first corrigendum to ISO 32000-2:2020 (PDF 2.0, second edition) was a major topic at TechConnect 2021. The corrigendum addressed ambiguities and errors discovered in the first four years of PDF 2.0 implementation by the developer community. Key areas addressed in the corrigendum: Associated files (/AF) clarifications — PDF 2.0 introduced /AF (associated files) for embedding structured data alongside PDF content (used in ZUGFeRD/Factur-X e-invoicing). The initial PDF 2.0 specification left some ambiguity about where /AF arrays could appear and what /AFRelationship values were valid in specific contexts. The corrigendum clarified these rules, resolving interoperability issues between different ZUGFeRD implementations. Transparency compositing group colour space — The rules governing the colour space of transparency groups were clarified, addressing cases where the group colour space conflicted with page-level output intent in a PDF/X context. Multiple implementations had diverged on the correct behaviour. Optional content state interactions — The interaction between /OCG state arrays and visibility expressions in complex optional content configurations had been interpreted differently by different viewers. The corrigendum specified the canonical behaviour. Page-level output intents — PDF 2.0 added page-level output intents, allowing different ICC profiles on different pages of a single document. The corrigendum refined the inheritance rules and interaction with document-level output intents. Interoperability test results — Sessions at TechConnect 2021 presented data from the PDF Association's interoperability testing programme, which had been running PDF 2.0 test files through multiple implementations to identify divergences. The test results drove the priorities for corrigendum content.

Cloud PDF processing: the pandemic-era transformation

The COVID-19 pandemic had produced an unprecedented acceleration in cloud-based PDF processing. TechConnect 2021 addressed the technical implications: E-signing volume surge — DocuSign's public filings showed that agreement volumes had more than doubled in 2020. Cloud signing platforms were processing PDF signature workflows at volumes that stressed infrastructure and revealed new failure modes: LTV signature generation at scale, cloud HSM certificate management, and the interaction between cloud-generated PDF signatures and enterprise document management systems. Remote notarisation — Online remote notarisation (RON) had been rapidly legalised across US states in 2020–2021 as pandemic emergency measures became permanent. RON workflows produced PDF documents with identity-verified signatures anchored to video recording sessions — a new document type that PDF processing systems needed to handle. Server-side PDF generation at cloud scale — Organisations that had processed PDFs locally were moving to cloud microservices. The technical challenges of cloud-scale PDF processing were discussed: Ghostscript and pdflatex containerisation; Chromium headless as a PDF generation backend; cloud function cold start latency for PDF operations; and the memory and storage characteristics of large PDF processing pipelines. Headless browser PDF generation — Chromium's print-to-PDF (via Puppeteer or Playwright) had become a widely used technique for HTML-to-PDF conversion. TechConnect 2021 addressed the quality and conformance limitations of this approach: Chromium-generated PDFs are not tagged, lack ToUnicode CMaps, and do not conform to PDF/A without post-processing. For accessibility-required or archive-required workflows, Chromium print-to-PDF was identified as unsuitable without remediation.

PDF/UA-2 early technical draft: developer preview

PDF/UA-2's first public committee draft was available in 2021, and TechConnect provided an opportunity for developer community engagement with the draft specification: The /B deprecation rationale — The removal of /B (bold) as a structure element was explained from the specification committee's perspective: /B had been widely misused as a visual styling tag rather than a semantic element, and the distinction between "this text is visually bold" (a presentation concern) and "this text is semantically strong/important" (a semantic concern) required clearer structure elements. /Strong (for semantic importance) and /Span with appropriate attributes (for visual styling) were the replacements. MathML integration design — The mechanism for embedding MathML in PDF was discussed in detail: using the /AF (associated files) mechanism with /AFRelationship set to /Supplement. Each /Formula element would reference a MathML file via /AF. Developers building scientific PDF pipelines discussed implementation feasibility and the toolchain implications for LaTeX, Word, and InDesign-based workflows. New namespace mechanism — PDF/UA-2 introduced a namespace mechanism for structure elements, allowing multiple standard tag sets (PDF 1.7 standard elements, PDF 2.0 standard elements, MathML, custom) to coexist in the same document without ambiguity. This was a significant architectural improvement over the role map approach used in PDF/UA-1. Developers attending TechConnect 2021 submitted feedback that influenced the final PDF/UA-2 specification. The iterative development process — draft → developer review → feedback → revision — was highlighted as a model for more effective standards development.

Frequently asked questions

What is the difference between PDF Days and PDF Association TechConnect?

PDF Days (Europe and North America) are broad-audience annual conferences covering all aspects of PDF: standards, implementation, business use cases, and industry trends. Attendees include software vendors, enterprise end users, records managers, accessibility specialists, and legal professionals alongside developers. PDF Association TechConnect is specifically targeted at software engineers and technical architects implementing PDF standards in code. Sessions focus on specification details, implementation challenges, interoperability test results, and developer tooling. TechConnect typically features more deep-dive technical sessions with code examples, specification analysis, and direct engagement with PDF specification committee members.

How does Chromium headless compare to dedicated PDF libraries for PDF generation?

Chromium headless (via Puppeteer or Playwright) generates PDFs by printing an HTML page to PDF — using Chromium's built-in print-to-PDF engine. This approach produces visually faithful PDFs from HTML/CSS input and is fast and simple to deploy. The limitations: (1) No tagged PDF output — the generated PDF is completely untagged, making it inaccessible and non-compliant with PDF/UA or PDF/A-a. (2) Missing ToUnicode CMaps — text extraction produces variable quality. (3) No PDF/A conformance — the output does not conform to any PDF/A profile without post-processing remediation. (4) Limited control over PDF internals — page size, margins, and print settings are accessible but PDF stream-level control is not. For accessibility-compliant or archival PDFs, dedicated PDF libraries (iText, pdfbox, pdf-lib, WeasyPrint) or purpose-built HTML-to-accessible-PDF tools are required.

What are associated files (/AF) used for in PDF?

Associated files (/AF) is a mechanism introduced in PDF 2.0 (and backported to PDF/A-3) for attaching related data files to a PDF document or specific objects within it. The /AF array on the document catalog (or on a page, structure element, or annotation) contains file specification dictionaries pointing to embedded file streams. The /AFRelationship key on each file specification indicates the semantic relationship between the PDF and the attached file. Common uses: ZUGFeRD/Factur-X e-invoicing (XML invoice embedded in PDF/A-3 with /AFRelationship /Data); PDF/UA-2 MathML (MathML fragments attached to /Formula elements with /AFRelationship /Supplement); scientific data (research datasets attached to PDF papers with /AFRelationship /Source); CAD files (engineering data attached to PDF drawings). The mechanism enables PDF to serve as a container format for structured data alongside the human-readable document.

Try it in your browser

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

Edit PDF Sign PDF Annotate PDF Open editor