Open yoavweiss opened 1 year ago
See this blog post. Tl;dr: there is no escape from clock drift. We can do all kinds of tricks to smoothen it, but in the end some of this would leak to the user. Suggesting to close this, and use variable time origin when necessary.
Following a discussion in the WG, it seems like we need to decide on the tradeoffs between consistency among different timeOrigins, and eventual skew of timeOrigins from the system clock (e.g. following system sleep).
Should implementations align the monotonic clock to the system clock when possible? (e.g. when initiating a new isolated browsing context)
/cc @noamr