An expensive query is run on every click about logseq-journals-nextprev

[isosphere]

Instead of "(pull ?p [*])" "(pull ?p [:block/journal-day])" might be sufficient.

I've benchmarked the * query with 100 samples, and for my journals it takes about 160 ms to execute the query.

I didn't have to compile 100 samples to see the difference; I'm immediately and consistently getting 80 ms with your suggested query. I'll implement it permanently, thank you!

A 50% improvement in execution speed is pretty good.

[YU000jp]

There's a callback for that.

logseq.App.onTodayJournalCreated(({ title }) => {

})

May need to update the version of the plugin API "@logseq/lib" from 0.0.15 to 0.0.17 for this to work normally. (package.json & yarm)

[isosphere]

May need to update the version of the plugin API "@logseq/lib" from 0.0.15 to 0.0.17 for this to work normally. (package.json & yarm)

I've done so, but it seems like the hook doesn't fire. Maybe I'm doing something wrong, or maybe "onTodayJournalCreated" refers to a specific type of journal creation (what's "today" mean here?). The hook isn't documented, unfortunately.

There doesn't seem to be a way to hook into deletion either, so maybe we lack the hooks required to reliably keep the cache updated. I'd rather the plugin behave reliably and a bit slow, than skip to pages that don't exist anymore or miss new ones but do so faster.

I wonder if the query would be faster if we didn't (pull ?p [*]) and instead picked and chose exactly what we wanted it to give us? I can try benchmarking that at some point, but the speed isn't bugging me yet with over a thousand journals.

[YU000jp]

Sorry. Two hooks caught fire, but it wasn't enough. I also have two years worth of journals accumulated, but the query speed still seems to be fine.

[YU000jp]

Instead of "(pull ?p [*])" "(pull ?p [:block/journal-day])" might be sufficient.

[YU000jp]

Great!