• onlinepersona@programming.dev
    link
    fedilink
    arrow-up
    3
    ·
    1 day ago

    What are you talking about? Logseq uses markdown. The database is built from the markdown files. You can regenerate it at any time.

      • pkjqpg1h@lemmy.zip
        link
        fedilink
        English
        arrow-up
        1
        ·
        21 hours ago

        But database version’s performance is great

        Benchmark graph: 4k movies Exported sqlite file

        Result compared to feat/db:

        app start time from 20s to 1 ~ 2s
        feat/db: loads 784571 datoms in 20077 ms
        this PR: restore-graph! loads 2880 datoms in 945 ms
        open "All pages" from 4s to 400ms
        open "#movies" from 5s to 400ms
        cmdk search && node reference are much faster, 3~4s to less than 100ms
        table view scrolling now maintains a minimum frame rate of >40 fps. Previously (in feat/db), such large table views were nearly unusable, with frequent frame drops to 1 fps during scrolling.
        add/remove object from "#movies" are much faster
        

        https://github.com/logseq/logseq/pull/11774