A single-process writer is reloading its own data

[mtrmac]

I’m currently observing that after a write to layerStore, reloadIfChanged triggers a reload. That seems unnecessary.

An unconfirmed hypothesis is that this is because we only update layerspathModified in load, so after a save, layerStore.reloadIfChanged always sees a timestamp update and triggers a reload.

[mtrmac]

Another, not really related issue: right now, we load the data when initializing a store, but that does not interact with lockfile.lw at all, so the next operation immediately triggers a reload. I suppose the initial store constructor should do a dummy _ = fooStore.modified()? OTOH that risks exacerbating #1420 .

[mtrmac]

I’m currently observing that after a write to layerStore, reloadIfChanged triggers a reload. That seems unnecessary.

An unconfirmed hypothesis is that this is because we only update layerspathModified in load, so after a save, layerStore.reloadIfChanged always sees a timestamp update and triggers a reload.

Confirmed now.

[mtrmac]

Cc: @alexlarsson — I’m probably not going to work on this soon; unlike me, you know if it makes a significant difference.