Paul Khuong
@pkhuong@discuss.systems
- 439 Followers
- 358 Following
- 1.7K Posts
@tobinbaker i don't see the event count for *writers*. for mutual exclusion, I'd expect a CAS spinlock (into futex), or maybe a ticket lock (coupled with the sequence).
@cfbolz I've been a little bothered (mostly theoretically) by the fact that allocations are sampled at most once, regardless of their size. Turns out the correction factor isn't that hard to compute.
An allocation of n bytes is sampled pn times in expectation. The probability of being sampled 0 times is (1 - p)^n, so the expected sampling count given the allocation is sampled at least once is pn / [1 - (1 - p)^n] (sanity check: reduces to 1 if n == 1)
@sree yeah! you have to make sure the program periodically calls the hot patched function/method instead of looping in the same stack frame, but it does work.
@sree yes, it kind of just works :) 'swhy i like to have an ipython listener in my python servers.
@pervognsen i think it's more likely that the FS just won't implement O_TMPFILE.
@bradlarsen @pervognsen turing award wen? ;)
@aw ggplot2 works for me when I want well supported stuff. never tried it off the happy path.
@hisham_hm @alerque @samueldr FWIW, I do rust without LSP all the time. I also decline to use libraries with API surfaces that feel impossible to explore from just the rustdoc.