Skip to contents

Acknowledgments

Source: ACKNOWLEDGMENTS.md

These packages stand on a great deal of other people’s work — code I import, data I serve, standards I follow, and research I build on. This file records the debts.

The work that pointed the way

The packages I learned from before writing my own:

  • urltools (Oliver Keyes, Jay Jacobs, Drew Schmidt) — the R package I reached for first, and the reason rurl exists.
  • hrbrmstr/punycode (Bob Rudis) — the R package whose puny_* / is_punycode() surface punycoder descends from before rebuilding it on libidn2 with explicit UTS #46 processing.
  • xsitemap (Florian Joly and contributors) — the prior R take on sitemaps that sitemapr measured itself against.
  • ultimate-sitemap-parser (Python) — the parser whose behavior sitemapr used as a parity oracle.

The people behind the projects I work on

These packages are an ecosystem — rurl, pslr, punycoder, pagerankr, and sitemapr — that lean on one another: rurl builds on punycoder and pslr; pagerankr and sitemapr build on rurl. My thanks go to everyone who has reviewed, tested, reported against, or otherwise helped shape any of them. The work is better for it.

What I rely on directly

The third-party libraries I import at runtime:

  • curl (Jeroen Ooms) — URL parsing and transfers, wrapping libcurl (Daniel Stenberg and contributors).
  • stringi (Marek Gagolewski and contributors) — Unicode text handling, on top of ICU (the International Components for Unicode).
  • xml2 (Posit) — XML parsing and XSD validation, wrapping libxml2 (Daniel Veillard and contributors).
  • httr2 (Hadley Wickham and Posit) — HTTP fetching, redirects, and retries.
  • htmltools, rlang, and tibble (Posit and the tidyverse authors) — report rendering, tidy evaluation, and the output row model.
  • igraph (Gábor Csárdi, Tamás Nepusz, and contributors) — the graph engine behind pagerankr, itself carrying PRPACK (David Gleich and others), ARPACK, and BLAS/LAPACK.
  • Rcpp (Dirk Eddelbuettel, Romain François, and contributors) — the R/C++ bridge in punycoder.
  • cpp11 (Davis Vaughan, Jim Hester, and Posit) — the R/C++ bridge in pslr’s matcher.
  • GNU libidn2 (Simon Josefsson and contributors) — the optional native Punycode backend in punycoder.

And the tooling that keeps all of this tested, documented, and honest: testthat, knitr, rmarkdown, withr, roxygen2, pkgdown, lintr, goodpractice, covr, cucumber, digest, air, and the oysteR / rosv vulnerability auditors.

The data I serve

  • The Public Suffix List and the volunteers at Mozilla and across the community who maintain it. pslr bundles a pinned snapshot as its source of truth for registrable domains, under the Mozilla Public License 2.0. See https://publicsuffix.org.
  • The Unicode Consortium, for the Unicode Character Database, the UTS #46 specification, and the IDNA conformance test data (IdnaTestV2) that punycoder is built and tested against.
  • The Sitemaps.org protocol and Google’s sitemap-extension documentation, which sitemapr’s bundled schemas model (clean-room, not copied).
  • jparise/chrome-utm-stripper and the AdGuard filter maintainers, whose published tracking-parameter lists seeded and sanity-checked rurl’s query denylist.

Standards and specifications

Much of this code exists to implement, or faithfully defer to, published standards. Some are cited by name in the source; others are only implied by the behavior. Both are recorded here.

URIs and URLs

  • RFC 3986 — Uniform Resource Identifier (URI): Generic Syntax.
  • RFC 3987 — Internationalized Resource Identifiers (IRIs).
  • RFC 2396 — URI: Generic Syntax (obsoleted by 3986, but part of the lineage the grammar traces back to). https://www.ietf.org/rfc/rfc2396.txt
  • RFC 1738 / RFC 1808 — Uniform Resource Locators and Relative URLs.
  • The WHATWG URL Standard — the living host-parsing and percent-encoding model rurl targets.
  • Web Platform Tests (web-platform-tests/wpt) — the WHATWG conformance corpus (urltestdata.json) rurl validates against.

Internationalized domain names and Unicode

  • RFC 3492 — Punycode, the Bootstring encoding for IDNA.
  • RFC 5890 / 5891 / 5892 / 5893 / 5894 — the IDNA2008 suite (definitions, protocol, tables, the Bidi rule, and rationale).
  • RFC 3490 / 3491 / 3454 — IDNA2003, Nameprep, and Stringprep, the superseded regime punycoder deliberately does not follow.
  • UTS #46 — Unicode IDNA Compatibility Processing.
  • UAX #15 — Unicode Normalization Forms (NFC).
  • UAX #44 — the Unicode Character Database.
  • UTS #39 / UTR #36 — Unicode Security Mechanisms and Considerations (referenced as scope boundaries, not implemented).
  • RFC 8753 — IDNA review for new Unicode versions.
  • RFC 3629 — UTF-8.

Hosts, DNS, and IP addresses

  • RFC 952 / RFC 1123 — host-name syntax (the STD 3 ASCII rules).
  • RFC 1034 / RFC 1035 — domain-name concepts and the label/length limits.
  • RFC 791 — the IPv4 dotted-quad presentation form.
  • RFC 4291 / RFC 5952 — IPv6 addressing architecture and canonical text representation.
  • RFC 6874 — IPv6 zone identifiers in URLs.
  • RFC 2606 / RFC 6761 — reserved and special-use domain names.
  • RFC 1918 / 6598 / 3927 / 4193 / 6052 — the private, shared, link-local, unique-local, and IPv4-embedded-IPv6 ranges behind sitemapr’s SSRF guard.

The Public Suffix List

  • The Public Suffix List format and prevailing-rule algorithm — the normal, wildcard, and exception rules, matched right-to-left with longest-match and exception-beats-wildcard precedence.

The research pagerankr implements:

  • Brin & Page (1998) and Page, Brin, Motwani & Winograd (1999) — PageRank and the random-surfer model.
  • Kleinberg (1999) — HITS (hubs and authorities).
  • Lempel & Moran (2001) — SALSA.
  • Gyöngyi, Garcia-Molina & Pedersen (2004) — TrustRank.
  • Haveliwala (2002) — Topic-Sensitive PageRank.
  • Kamvar, Haveliwala & Golub (2004) — PageRank convergence.
  • Langville & Meyer (2004) — Deeper Inside PageRank.
  • Boldi, Santini & Vigna (2005) — PageRank as a function of the damping factor.

Sitemaps, crawling, and documents

  • The Sitemaps.org Protocol 0.9 — the core <urlset> / <sitemapindex> format and its limits.
  • Google’s sitemap extensions — Image 1.1, News 0.9, Video 1.1, PageMap 1.0, and the xhtml:link hreflang convention.
  • RFC 9309 — the Robots Exclusion Protocol (the Sitemap: directive).
  • RFC 9110 / 9111 — HTTP semantics and caching.
  • W3C XML 1.0, XML Namespaces, and XML Schema 1.0 (XSD).
  • The W3C Date and Time Formats profile of ISO 8601 — lastmod parsing.
  • ISO 639, ISO 3166-1 alpha-2, and BCP 47 (RFC 5646) — the language and region codes referenced by the sitemap extensions.
  • RFC 1952 — the gzip file format.

R, CRAN, and the compiler ecosystem

Thanks to the R Core Team, whose language and base facilities (utils, tools, the connection and compression layers) carry all of this; to the CRAN volunteers, whose standards and checks make a package trustworthy; and to the C and C++ communities behind the native backends — libcurl, ICU, libidn2, libxml2, PRPACK/ARPACK — and the Rcpp and cpp11 bridges that let R reach them.

AI-assisted development

These packages were built with the help of AI development tools. Claude (Anthropic), through Claude Code, was the primary development assistant and is credited as co-author across the commit history; Codex (OpenAI) served as a design reviewer and caught real errors in the harder specifications. The design decisions, the standards conformance, and the responsibility for what ships are mine.

Alternatives

If one of my packages isn’t the right fit, here’s other work worth knowing about.

Instead of rurl

Instead of punycoder

Instead of pslr

Instead of pagerankr

Instead of sitemapr

A note on licenses

This file is about gratitude, not legal terms. The binding third-party license and attribution notices live alongside the code in each package — see that package’s THIRD_PARTY_NOTICES.md, inst/NOTICE, and LICENSE files. In particular, the Public Suffix List data bundled in pslr is licensed under the Mozilla Public License 2.0, separately from the MIT-licensed package code.