why would you call this a "pet peeve" it's obviously a matter of education not of people being assholes how tremendously pretentious.

d@nny mc²

why would you call this a "pet peeve" it's obviously a matter of education not of people being assholes how tremendously pretentious. i learned something but i'm terribly upset about it https://fosstodon.org/@djspiewak/113600799130927494

d@nny mc²

also "no point" is incredibly strong language. obviously the people who did this thought there was a point to doing it more slowly

d@nny mc²

note heavy use of jargon but no links or citations

d@nny mc²

pet peeve: when people didn't attend the same university as me

Asta [AMP]

@[email protected] I like that the first response is "random jitter is an inherent property of the system" (which is almost certainly correct given the possibility of delays in network transmission; whether the magnitude of that stochastic element is large enough is another story)

and his response is "That jitter is a constant factor though" which... look. I'm not a network engineer. My understanding of physics tells me there is likely going to be a noise factor in transmissions times pulled from a Gaussian. If that's correct, then... that's not a 'constant factor'. "the jitter is a constant factor" ?? is meaningless??

Asta [AMP]

@[email protected] Perhaps someone with a deeper understanding of network transmission physics and real world scenarios can chime in for me here.

Again, I would think that if anything, it would just be a matter of the stochastic element being insubstantial relative to what is necessary?

d@nny mc²

@aud i think the duration of the request might be expected to be longer than the jitter induced by unreliable clocks and network transmission times

d@nny mc²

@aud in any case it's not clear from the screenshot that there's not random jitter being applied making all of this moot lmao

Asta [AMP]

@[email protected] fair! I sort of assumed that was gonna be the case based on what he was saying, but effectively decrying a choice as pointless and then saying "that jitter is a constant factor" is... I mean, that's just not correct.

Unless he means it's a variable that is always present, but since he seems to be such a stickler for rules...

Asta [AMP]

@[email protected] lmao true. It's a power series based on 3 but that doesn't mean it's necessarily accurately reporting the times (it could say 3^n and just not be reporting the jitter) or the request duration could be within the 1 second range so it just gets rounded out or whatever.

d@nny mc²

@aud yeah i decided i'm not gonna interact but this guy was exactly as annoying on twitter where i have him blocked

d@nny mc²

@aud the jitter of network transmission probably has elements of true over artificial randomness too

Asta [AMP]

@[email protected] lmao fosstodon gonna brosstodon

d@nny mc²

@aud they have a no shitposting policy

Asta [AMP]

@[email protected] for sure. There's also theoretically jitter introduced by system load and other such factors but I would imagine that if this type of stuff is running on not-overloaded systems designed for network ops it's probably even less than transmission randomness.

but what, you don't trust the famously backdoored /dev/urandom? What was the CPU entropy source that famously the NSA had backdoored?...

Asta [AMP]

@[email protected] really? Because 95% of their posts are shit.

d@nny mc²

@aud NSA backdoored the clipper chip and then DUAL_EC_DRBG thanks to RSA through project bullrun, i wasn't aware of cpu entropy?

d@nny mc²

@aud dev urandom works great i was just noting a fun fact about sources of randomness although i may have actually been completely wrong here

Asta [AMP]

@[email protected] It's not confirmed, but there were some theories. Apparently this was like, a decade ago though:

https://security.stackexchange.com/questions/42164/rdrand-from-dev-random

Basically, RDRAND from Intel was potentially compromised, or there was concern it might have been. I don't remember all the exact circumstances. I think urandom was the one you should use, or something. I can't remember.

Asta [AMP]

@[email protected] no no, I think you're right. /dev/urandom is awesome.