euandre.org, branch main

Stop claiming music pages

2026-09-12T21:43:22Z

Every .ly is in the untracked STUFF/, so no music page is built. mkdeps.sh keeps its rules for when one is --- it is the README that was describing machinery as though it were content. Co-Authored-By: Claude Opus 5 (1M context) Claude-Session: https://claude.ai/code/session_01KxLpaNQThKfMG3ecCJptP9

Five languages, not six

2026-09-12T17:30:09Z

German went with po/de.po in 6ca718a; what is left is English and the four po4a writes from it --- Esperanto, Spanish, French and Portuguese. The count sits in the NAME line every page carries, so it was wrong in eight places at once. Co-Authored-By: Claude Opus 5 (1M context) Claude-Session: https://claude.ai/code/session_01KxLpaNQThKfMG3ecCJptP9

Compute each selector once, not seven times

2026-09-12T17:15:16Z

mkdeps.sh asked the same questions over and over: "pages" is built on four of the other selectors, the tagged list mkwb reads asks for all seven again, and "collections" alone was computed nine times, a find and a grep and a sort apiece. It spawned 372 processes to print 2450 lines. Every answer is now computed once, up front, into a file named after it, and read back from there. The three hand-rolled caches go with it: they were the same idea, done three times, and the loop that fills them says the dependency order out loud. Two smaller things fell out on the way: collections ran dirname once per collection where sed does it in one, and articles ran one find per collection where find takes them all at once. The order the loop fills them in is not decoration. A pipeline's stages run side by side, and only_masters and with_translations are two stages of one selector that both read what po4a writes --- so an answer computed where it is first needed can be read while it is still half-written, which is a race that shows up as an empty list and a deps.mk missing half the site. Filled in order, up front, nothing is ever read while it is being written. 372 processes to 109, 0.47s to 0.24s, byte-identical deps.mk over a dozen runs. The floor now is exec(2): the tools this reaches for cost 1.1 to 2.5ms each to start, and a hundred of them is the whole of the remaining runtime. Co-Authored-By: Claude Opus 5 (1M context) Claude-Session: https://claude.ai/code/session_01KxLpaNQThKfMG3ecCJptP9

Carry a long-description

2026-09-12T16:55:56Z

pristineg required one as of its 5fa9369: a catalogue reproduces both files verbatim, and the reader who wants more than "Blog and personal website (euandre.org)." had nowhere else to go. Co-Authored-By: Claude Opus 5 (1M context) Claude-Session: https://claude.ai/code/session_01KxLpaNQThKfMG3ecCJptP9

Declare the site's language, and name the tools as the renames left them

2026-09-12T12:28:31Z

lang is "html": what this builds is a tree of pages, and pkgs reads that to pick mkwbpkg, which adds the gzip phase that puts a twin beside every file that shrinks. The documentation catches up too. The rename sweep matched "mkwb(1)" but not "*mkwb*(1)": with the section in the brackets after the closing star, the name is not what the paren follows. Co-Authored-By: Claude Opus 5 (1M context) Claude-Session: https://claude.ai/code/session_015bWfjj7PP24iF5MqFHmdeC

deps.mk: Regenerate, for the tool names the rename moved

2026-09-12T12:11:05Z

The file is mkwbg(1)'s output, and it was still the old mkwb's: four rules read a value back with "$(mkwb get ...)" inside a command substitution, which the build could not find. sh: line 1: mkwb: command not found make: *** [deps.mk:2186: ...updatedat-check] Error 127 mkwbg already prints its own name; this is only the copy here catching up. Co-Authored-By: Claude Opus 5 (1M context) Claude-Session: https://claude.ai/code/session_015bWfjj7PP24iF5MqFHmdeC

meta.txt: Declare the imagemagick the favicon rules run

2026-09-12T12:07:39Z

mkwbg(1) emits two rules that turn src/content/img/favicon.svg into favicon.ico and favicon.png, and both call convert(1): sh: line 1: convert: command not found make: *** [deps.mk:2037: src/content/favicon.ico] Error 127 The rules are not new --- they were in the tree before the renames, and a build sandbox has never had imagemagick. What changed is that the build now reaches them: it used to stop earlier, on the mkwb that the rename had turned into mkwbg. Co-Authored-By: Claude Opus 5 (1M context) Claude-Session: https://claude.ai/code/session_015bWfjj7PP24iF5MqFHmdeC

m

2026-09-12T03:15:09Z

Drop gettext from meta.txt

2026-09-11T17:21:27Z

Nothing here runs msgfmt, or any other gettext tool: this site has no message catalogue of its own, and the wording tables it does translate are committed. The entry was asking for a package the build never opens. Co-Authored-By: Claude Opus 5 (1M context) Claude-Session: https://claude.ai/code/session_01BMowJCogxLWcMdVYYhbLpH

Say in meta.txt what the package actually asks for

2026-09-11T16:52:18Z

Nothing reads this file here --- the site is not in pkgs' repos.txt, so its package is hand-written rather than generated from this --- and it had drifted from that package in both directions. Out: lilypond, for which there is no source (mkdeps.sh says as much: "There is no music yet"); ruby-asciidoctor, which nothing invokes; and tar, which was listed twice, once as a sibling repository and once as a system package. In: eslaides, misct, mktorrent and mkwb, which the site is built with, and ghostscript, gnupg and moreutils, which it runs. It now mirrors the entry in pkgs exactly, so the two can be compared by eye. Co-Authored-By: Claude Opus 5 (1M context) Claude-Session: https://claude.ai/code/session_01BMowJCogxLWcMdVYYhbLpH