Till Schneidereit (Aug 25 2025 at 11:46): Till Schneidereit (Aug 25 2025 at 11:46):@Tomasz Andrzejak what I said about using Unix timestamp 0 as the timeOrigin in StarlingMonkey unfortunately was wrong :frown: What I didn't account for is that this could result in violations of monotonicity, for example if the system clock is changed.
I think we do need to employ the kind of scheme I briefly sketched out: during wizer resumption, check if a timeOrigin was set during wizening. If so, take a "second origin", and subtract the delta between that and the first one from all values. The result should I think be that it seems like no time passed between wizening and resumption, and we retain monotonicity.
Last updated: Jan 10 2026 at 20:04 UTC