summaryrefslogtreecommitdiff
path: root/README.org
diff options
context:
space:
mode:
authorcashmere <cashmere@autistici.org>2026-03-10 12:10:44 +0100
committercashmere <cashmere@autistici.org>2026-03-10 12:10:44 +0100
commit007e9081b5333ae5fa91b22d85d7c3359b3e5e12 (patch)
treea23eab6788fb554ea44adbce4092635a755268c7 /README.org
parente8f259d8c03126d548ec28dd9a4bcc41096ab9f9 (diff)
downloadkitty-graphics-007e9081b5333ae5fa91b22d85d7c3359b3e5e12.tar.gz
experiment: add pre-redisplay-functions as alternative refresh triggerexperiment/pre-redisplay-functions
Add feature flag (kitty-gfx-use-pre-redisplay) to compare post-command-hook vs pre-redisplay-functions as the scheduling trigger for image refresh. Includes instrumentation counters (trigger count, skipped count) logged per refresh cycle for quantitative comparison. Results: pre-redisplay-functions fires ~36x more often than post-command-hook for identical user activity (3176 vs 89 refresh cycles in comparable sessions). Emacs redraws far more often than users run commands (cursor blink, timers, internal redraws), and pre-redisplay-functions fires per-window on every redisplay. The current post-command-hook approach remains superior.
Diffstat (limited to 'README.org')
0 files changed, 0 insertions, 0 deletions