Skip to content

Latest commit

 

History

History
41 lines (28 loc) · 2.71 KB

TODO.md

File metadata and controls

41 lines (28 loc) · 2.71 KB

TODO

Open issues and PRs as of 2026-09-09.

Pull requests

  • #95 — Bump brace-expansion 1.1.15 → 1.1.18 (dependabot, mergeable, clean).
  • #94 — Bump js-yaml 3.14.2 → 3.15.2 (dependabot, mergeable, clean).
  • #41 — WIP scripting. Draft, still mergeable. Adds intro prose and document.write() description to scripting.md, but leaves an empty ![]() for the HTML standard's parser diagram, a "TODO discuss nested document.write limit", an empty "The scripting flag and the noscript element" section, and the existing bare TODOs.
  • #24 — Expand on srcset and sizes. Conflicting; base predates the current ### Sizes section in microsyntaxes.md, so it needs a rebase. Also has 5 unresolved TODOs (CSS value syntax link, media condition vs media query rationale, whether width was dropped, min()/max()/clamp() dates and browser support, a min/max example).

Security chapter

security.md is still three bare TODOs, so most of these are unwritten rather than incomplete.

  • #19 — Explain how changes to the HTML parser can cause XSS problems.
  • #76 — Hoisting attributes causes DOM clobbering with getElementById (PortSwigger service worker hijacking research).
  • #1 — Use the Google noscript XSS as an example.
  • #55 — Trusted Types.
  • #56 — Sanitizer API.
  • #74 — CWE over time (NVD visualization; owner later added a 2025 MITRE/CISA "XSS ranked #1" link in the thread).

Parser and spec content

  • #84 — Reader wants detail on special tags: what "namespace boundaries" meant in the 2008 whatwg post, why exactly mi/mo/mn/ms/mtext are MathML text integration points, and the rationale for the element set in "has an element in scope".
  • #45 — Document document.execCommand with insertHTML under "Other parser APIs" in scripting.md. The linked Trusted Types issue is also a case study for the security chapter.
  • #44 — Write something about declarative shadow DOM.
  • #37 — Add a section about XML declaration character encoding, near "Bogus comments".
  • #42 — Clarify SGML vs WebSGML (ENR+WWW) for />. The book's claim that <link ... /> and <link ... >> are equivalent is disputed; the 1998 WebSGML NETENABL / IMMEDNET features are the relevant part. Verify before rewriting.

Clarity, from reader confusion

  • #29 — Nested forms confused a reader over which element node and "current node" refer to. Owner's own conclusion: spec quotes in the book should carry links and mark up variables so it's clear what's a variable.

Implementations chapter

implementations.md is a bare TODO.

  • #22 — Servo's HTML parser.
  • #23 — Cloudflare's LOL HTML rewriter.
  • #25 — Flow browser (quirksmode interview: the parser was easier than expected thanks to the spec).