Commits for braney

1f88a89d5bb7bada7c52ad89f910c6796d1340ed Tue Jun 2 12:26:10 2026 -0700

Add note on quickLifted CCDS pages explaining the data is from the source assembly, refs #36125 When a user clicks the CCDS link on a quickLifted gene track, the CCDS page necessarily loads from the source assembly (hg38), since CCDS tables exist only there. Mark the click with a one-shot quickLiftCcds CGI var carrying the destination assembly, and on the CCDS page print a note explaining that the CCDS gene model and coordinates shown are from the source assembly, not the destination assembly the user was browsing. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- src/hg/hgc/ccdsClick.c - lines changed 17, context: html, text, full: html, text
- src/hg/hgc/hgc.c - lines changed 1, context: html, text, full: html, text

874066eeba199bb09909de98bb8ba2d75b1b019c Tue Jun 2 18:10:38 2026 -0700

Cache FreeType glyph advances and rendered bitmaps for faster label drawing Drawing many long labels was dominated by FreeType reinterpreting each glyph outline (FT_Load_Char) on every call, both to measure widths and to render; profiling showed ~3/4 of the glyph loads were width measurement alone. Cache the advance and the rendered GRAY coverage bitmap per (pixel size, character). The cache hangs off face->generic.data (freed via face->generic.finalizer), so it is private to each FT_Face and needs no locking once the draw is multi-threaded with a per-thread face. Caching covers single-byte chars on the default identity-transform text path (ftText); ftTextInBox (transformed/scaled text) and Unicode fall back to a direct FT_Load_Char. Output is pixel-identical to before (verified AE=0/MAE=0 for both bitmap and FreeType fonts); roughly halves the FreeType render time on label-heavy views. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- src/lib/freeType.c - lines changed 151, context: html, text, full: html, text

07e5a2c08f148d8bf31219af8c2218cd2cc893cf Wed Jun 3 09:16:15 2026 -0700

hgTracks config: pair FreeType font names with their files in one table The configure-page font list lived in two parallel arrays, freeTypeFontNames[] and freeTypeFontFiles[], matched only by position. Nothing tied a name to its file, and maybeNewFonts() looked a name up in the first array and then indexed the second with no bounds check -- so any drift in order or count between the two, or a selected name that matched nothing, silently rendered the wrong font or read past the end of the array. Merge the two arrays into a single struct freeTypeFont[] of {name, file} rows so the two can no longer get out of sync, and make maybeNewFonts() fall back to the bitmap engine when the selected font isn't one we know instead of indexing out of bounds. Same fonts and same name-to-file mapping as before. Also fix faceNames[] to size by element count (ArraySize) rather than byte count (sizeof). refs #37698 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- src/hg/hgTracks/config.c - lines changed 65, context: html, text, full: html, text

60d8399ce8617bf31d27b7752576b66a9f29d9b2 Wed Jun 3 09:20:22 2026 -0700

hgTracks config: document how to add a new FreeType font Add a "HOW TO ADD A NEW FONT" comment block above the freeTypeFonts[] table explaining the two steps (place the file under freeTypeDir, add a {name, file} row), how freeTypeDir resolves relative to the CGI working directory (so it finds the shared htdocs/urw-fonts even from a per-user sandbox), the name/face/style convention used by the configure-page dropdown, and the gotchas -- notably that variable fonts like InterVariable.ttf render only their default master instance, so a static single-weight file should be used instead. refs #37698 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- src/hg/hgTracks/config.c - lines changed 36, context: html, text, full: html, text