Max reconstruction delay as a function of slot time #8067
Conversation
PoulavBhowmick03
force-pushed
the
delay/slot_duration
branch
from
25ab98c to
8012a09
Compare
| } | ||
| InboundEvent::Msg(DelayColumnReconstruction(request)) => { | ||
| let mut reconstruction_delay = QUEUED_RECONSTRUCTION_DELAY; | ||
| let slot_duration = self.slot_clock.slot_duration().as_millis() as f64; |
There was a problem hiding this comment.
We avoid doing any float arithmetic due to rounding issues and general flakiness. Integer arithmetic in milliseconds would be my preference
There was a problem hiding this comment.
yep, might be worth adding this preference to CLAUDE.md too
There was a problem hiding this comment.
We avoid doing any float arithmetic due to rounding issues and general flakiness. Integer arithmetic in milliseconds would be my preference
got it. will make that change
There was a problem hiding this comment.
We avoid doing any float arithmetic due to rounding issues and general flakiness. Integer arithmetic in milliseconds would be my preference
in that case, i was initially having RECONSTRUCTION_DEADLINE as a fraction (1/4) and then multiplying the slot_duration in millis with the numerator and denominator of the same. should i keep that approach?
let reconstruction_deadline_millis =
(slot_duration_millis * RECONSTRUCTION_DEADLINE.0)
/ RECONSTRUCTION_DEADLINE.1;
let reconstruction_deadline = Duration::from_millis(reconstruction_deadline_millis);
There was a problem hiding this comment.
I think the simplest is probably something like slot_duration_ms * reconstruction_deadline / 100. Or just divide by 4. Seems a bit unnecessary to multiply by one
michaelsproul
added
the
waiting-on-author
PoulavBhowmick03
force-pushed
the
delay/slot_duration
branch
from
8012a09 to
9d896af
Compare
Issue Addressed
Fixes #8054