an opinionated terminal multiplexer in C with programmable pane chrome https://git.sl0p.foo/slosh.git/atom/?h=v0.1.5 2026-09-02T14:26:29Z 2026-09-02T14:26:29Z user user@clank 2026-09-02T14:26:29Z urn:sha1:920a222cc9d135bdbd7d2174b3257c530e9d53c5 2026-09-02T14:24:41Z user user@clank 2026-09-02T14:24:41Z urn:sha1:c58a875ef3bcb464cdbea72bdbd0a874b6f93729 The lookup miss now writes the embedded entry to ~/.terminfo and uses it, so a bare box gets the full TERM with no step at all. An entry that exists anywhere is respected -- possibly newer, certainly somebody's -- and a HOME that cannot be written leaves the xterm-256color fallback standing. --install-terminfo stays as the by-hand spelling; the writer moved to pty.c where the panes are. 2026-09-02T14:22:03Z user user@clank 2026-09-02T14:22:03Z urn:sha1:f9986df0da09cc692e168e33ef9ccfab5b9d2faf The xterm-256color fallback works everywhere; the real entry is better, and a tarball user on a bare box had no easy way to get it. The entry is generated from the vendored tree (contrib/terminfo, so it describes the terminal core we actually ship), compiled with tic, embedded as bytes, and written to ~/.terminfo under every name it answers to, letter and Darwin-hashed layouts both. No root, no package, no conflict with a ghostty installed later. Proven end to end in a mount namespace: bare box falls back, installs, and the next pane says xterm-ghostty. 2026-09-02T14:15:55Z user user@clank 2026-09-02T14:15:55Z urn:sha1:07194e139fbb90cb946be4d1e50e4937dab1ce04 TERM=xterm-ghostty was unconditional, but the entry ships with ghostty and a machine that has never seen ghostty has ncurses' database and nothing else. There, nano exits the moment it opens (C-a e flashing a pane) and a shell that cannot look up kbs guesses at backspace. The entry is now looked up the way curses would -- TERMINFO, ~/.terminfo, TERMINFO_DIRS, the system dirs, letter and Darwin-hashed layouts -- and when absent, panes get xterm-256color, which every ncurses database carries. Verified both ways with a mount namespace hiding the database. 2026-09-02T14:10:48Z user user@clank 2026-09-02T14:10:48Z urn:sha1:9cc6ff911ac794c5831407358fddb7b7c9c24ef6 The v0.1.4 arch packages introduced themselves as 0.1.4-dirty: the tag's tarball has no .git, makepkg builds it inside the release box's repository, and git describe walked up and answered for the outer tree -- mid-release, when the paperwork is honestly dirty. Found at 0.1.2, diagnosed, and then left unfixed by yours truly. Two fences now. contrib/version only trusts a .git whose toplevel is its own root (helps every future tarball). The PKGBUILD exports GIT_CEILING_DIRECTORIES for build and package (helps regardless of which contrib/version the pinned tarball carries, which is why the first rebuild still said -dirty: the fix was outside the tarball). pkgrel=3; both architectures rebuilt, verified 'slosh 0.1.4', and republished. 2026-09-02T13:58:12Z user user@clank 2026-09-02T13:58:12Z urn:sha1:60d622062aefa1258e348c4085ab33e0e7e2ae48 Steps 6 and 7 rewrite the formula and PKGBUILD; building the site from the tag checkout with those uncommitted made git describe say -dirty, and v0.1.4's demo introduced itself as 0.1.4-dirty+webdemo. Commit the pins first, build the site from a clean tag second. 2026-09-02T13:57:00Z user user@clank 2026-09-02T13:57:00Z urn:sha1:0280bc0635cb02ff8a96be7f5d730fee9161c307 2026-09-02T13:47:10Z user user@clank 2026-09-02T13:47:10Z urn:sha1:975a8933685a058147a2fc2546ecddc74a6f2d91 2026-09-02T13:45:32Z user user@clank 2026-09-02T13:45:32Z urn:sha1:311979f10132697b990a670c6c99b19c1c33d623 Three weeks of upstream: 504 commits, 120 touching the terminal core -- OSC 5522, kitty relative placements, C0 fixes, the wuffs pixel path. Rebuilt from ghostty's own dist step (zig build dist -Demit-lib-vt=true), zig floor unchanged at 0.16.0, and slosh's C usage compiled unmodified. Patch 0001 is re-implemented, not retired: upstream rewrote the area into ImageStorage.clearScreen but kept the behaviour we patch out -- a terminal clear still reclaims transmitted-but-unplaced images, which breaks the upload-once place-per-frame pattern the protocol's two verbs exist for. The patch now removes just that sweep; the placement clearing and orphan reaping stay upstream's. Their recent 'preserve data on unmatched delete' fixes the same class of bug, so sending ours upstream remains the way to actually retire it. 2026-09-02T13:38:38Z user user@clank 2026-09-02T13:38:38Z urn:sha1:8c1553801a9812bf4e8541872eb720470ce89a0b Vendoring by pinned commit stays (D12) -- the package channels build from source archives where submodules are empty, the arch build is offline-clean because the core is in the tarball, and upstream has no stable release channel to depend on anyway. What was missing is the watching: vendor-check reads the pin from vendor.json, asks the GitHub API what happened since, and lists the commits touching the terminal core. First run found 120 relevant commits in three weeks, including upstream fixing the same class of kitty-graphics bug our patch 0001 fixes -- which is the argument for finally sending it.