> The upstream release notes include for 0.9.75 'The main improvement is
> the implementation of workaround for some OSes (like OpenBSD 7) where
> "monotonic" clock may jump back.' I don't recall anything reported to
> bugs@ about this and that is clearly incorrect behaviour, any idea
> what's going on with that?
This was observed a few times over the years, but nobody ever made an
actionable report, as far as I know. For example there used to be a
workaround in Rust for this:
https://lobste.rs/s/eo82wo/monotonic_clocks_are_not_monotonic
which points to
https://github.com/rust-lang/rust/issues/48514
which is useless.
The commit of the libmicrohttpd workaround states it as a fact that
OpenBSD's monotonic clock may jump backwards up to 500ms, which seems
quite a lot...
https://git.gnunet.org/libmicrohttpd.git/commit/?id=73c43a7ef7bdccda1b6b6714d2d64b1714e9cdad
No comments:
Post a Comment