tirclogv, branch masterAn IRC log viewerhttps://git.tomsmeding.com/tirclogv/atom/?h=master2026-05-10T20:20:24ZAdd tirclogv.prof to .gitignore2026-05-10T20:20:24ZTom Smedingtom@tomsmeding.com2026-05-10T20:20:24Zurn:sha1:08e042b949ca358a86c256d137379e76f3881bfcClean up2026-05-10T20:20:13ZTom Smedingtom@tomsmeding.com2026-05-10T20:20:13Zurn:sha1:da023dfdc4884325fb62b3b101fcc8ea44772752meta-externalagent, stop crawling all the ?eid= URLs2026-05-10T20:05:54ZTom Smedingtom@tomsmeding.com2026-05-10T20:05:54Zurn:sha1:e0e6b516b9dd132e067a226ff7fdf56d3e556559style.css: Add unline to <a> also on :focus2026-05-10T17:39:22ZTom Smedingtom@tomsmeding.com2026-05-10T17:39:22Zurn:sha1:50e74b214133adddf252a7fb7962272508d6d03dMigrate index and calendar templates to new renderer2026-05-10T17:38:59ZTom Smedingtom@tomsmeding.com2026-05-10T17:38:59Zurn:sha1:ad7ce6aacdb95f9bdf40a566186f5e2a220bdb5eGenerate renderer functions using TH2026-05-10T16:54:45ZTom Smedingtom@tomsmeding.com2026-05-10T16:54:45Zurn:sha1:5d6f1f42eec48ae1671e54cdd127f39318416498
These take (and project from) a data type instead of hash maps, _and_
use text-builder-linear, _and_ have the full template inlined. The
result is that this is the fastest renderer yet.
Rendering performance experiments2026-05-09T22:37:10ZTom Smedingtom@tomsmeding.com2026-05-09T22:36:36Zurn:sha1:66e9a4f242b9f02a7fcf6b5fc610417a50e1ba87Less warnings2026-05-09T22:36:58ZTom Smedingtom@tomsmeding.com2026-05-09T22:36:58Zurn:sha1:a0ece65727292ed5a9dc165686f5ec5f891ab30cMake genEventID 5x as fast2026-05-09T22:34:57ZTom Smedingtom@tomsmeding.com2026-05-09T22:34:57Zurn:sha1:b8c4db39334c6612bd966043d73eb8c1fcacb5b8
showIntAtBase wasn't inlined, so not specialised to Word64, so all the
arithmetic went via typeclasses (inluding a number of fromIntegral
@Word64 @Int, which went via Integer).
Optimisations to log rendering with mustache2026-05-09T10:49:00ZTom Smedingtom@tomsmeding.com2026-05-09T10:48:36Zurn:sha1:a63a096d075797f12ed614d5029e16f55a8bc769