03:40 <dnkl> aelius: foot isn't emitti6 the events. the compositor is

04:07 <aelius> hrm

10:56 <dnkl> did a performance benchmark of 10-bit surfaces in foot; a bit over 2x slower... guessing packing an unpacking 10-bit numbers is slow in pixman

18:05 <aelius> I read the performance overview readme once and didn't internalize it much. From my rather end user perspective, foot is "just correct in the first place" as I swap terminals around my tiling wm, whereas wezterm or ghostty will have their windows moved, but the contents notably "shift" a frame or so later

18:07 <aelius> However, when viewing an emoji-test.txt file I found, foot rather struggles to scroll through it, whereas the others don't

18:08 <aelius> I assume that's in-line with what's expected

18:09 <aelius> This one https://unicode.org/Public/emoji/15.1/emoji-test.txt

18:29 <dnkl> aelius: loading an emoji is relatively slow. Once all emojis have been cached, it will render *much* faster

18:29 <dnkl> you may have to scroll back and forth a couple of times to load all of them

18:30 <dnkl> I think there's also a tweak option in foot to speed it up, at the expense of worse rendering quality. I can look that up tomorrow

21:58 <aelius> yeah it does fix right up after caching. But there's no real world scenario where I'd see that many emojis anyway.