Show newer

@msuriar

A contrary opinion: it's terrible. You can do whatever you can do with it with subqueries, any reasonable optimizer will treat both identically, and HAVING makes things appear less composable than they are (tbf group by by itself already has that property, but knowing about HAVING but not about subqueries makes one assume that SQL is noncomposable).

@neilk

What's the typical sequence of steps if (a) no one volunteers why (presumably because no one knows) (b) when ~half the people think it's obviously self-explanatory and the other ~half think there's no good reason?

robryk boosted

We have a wonderful Slack channel called #why which is filled with questions like "why do we do <thing>" or "does anyone remember why we do <thing>"

1. It's kinda helpful when you scale up staff by 100% in approximately one year

2. Even the CEO asks why questions!

3. At year 6 or so of a startup there are lots of decisions to be reconsidered

robryk boosted

Some people, when confronted with a key management problem, think “I know, I'll use a Public Key Infrastructure.” Now they have a Problem Authority.

@eta Do you know what made that toot cursed? (And which one was it?)

@rq @starshine unless your serial connection is fast enough to support e.g. vi and everything knows which escape sequences to use.

@niconiconi You mean you'd get 3 of them per word? You could fit 5 12-bit values in each 64-bit entry, but that's not that much of a gain over 4 16bit values~~

@lauren Interesting. I never considered that earthquakes can cause more noticeable rotations than accelerations (I guess the noticeability thresholds might not be intuitive).

@lauren What's a rolling earthquake (sadly, the MTG card of that name makes looking it up nontrivial)?

@mjg59 Ah, I see. Indeed this is how one often sees posts from people one does not follow.

(One of my gripes with how APub is used by all(?) Fedi software is that they try to make it impossible to interact with the boost itself, and try to redirect all such interaction to the boosted message.)

@mjg59

Unless you (or someone else on your instance) actually follow that person, in which case replies should have been forwarded; or am I wrong?

@rq Do you mean clauses preventing commercial use? If not, then I'm really curious what kind of clauses you meant.

@grrrr_shark

Hm~ this made me realize that I don't really know whether acquiring one autoimmune problem makes acquiring other more likely (and I wonder how one can try to tell, given that likely there are confounders that are hard to notice and that make acquiring autoimmune issues more likely).

@grrrr_shark

From what I know, the most important failure mode of rapid tests (and to a lesser extent ~any tests) is that they will yield a very large fraction of false negatives early on. Do you know any estimates of repeated false negative probability over 2-3 tests over 2-3 days (i.e. probability of 2-3 tests being negative over 2-3 days while sick with COVID and symptomatic)?

When I tried to estimate it most recently, I ended up finding things that suggested wildly different rates of singular false negatives across different test manufacturers (by more than 2x IIRC), but sadly couldn't find anything about the repeated testing setup (I baselessly suspected that it would decrease the inter-manufacturer spread).

@markstos Why is the total kinetic energy of the car relevant?

robryk boosted

Szukam pomysłu, jaką pracę mógłbym wykonywać 

@mgorny

> Nie mam też doświadczenia jako sysadmin.

Czy nie jest tak, że niedoszacowujesz? Przynajmniej w okolicach bardziej devopsowych istotne rodzaje doświadczenia to często takie, które pozwalają (a) stwierdzić co jest nie tak z tym serwerem (b) oautomatyzować sobie zarządzanie większa liczbą takowych. Strzelałbym, że typowy maintainer Gentoo ma co najmniej trochę doświadczenia z (a) (no, może niekoniecznie z wersją typu "czemu I/O jest tu takie wolne", ale dużo z wersją typu "czemu temu ustrojstwu nie podobają się nasze biblioteki dynamiczne/czemu tu działa podczas gdy tam nie działa"). Naprawianie systemów budowania powinno dawać doświadczenie pomagające z (b): jak już widziałeś ileśtam rzeczy, które się zepsuły, bo wszystko było straszną gulgutierą, nabierasz intuicji pomagających wykrywać gulgutierowate pomysły i ich unikać.

Gdy przeprowadzałem rozmowy o pracę w okolicy devops (disclaimer: potencjalnie dość nietypowe), najbardziej się przejmowałem tym, czy potencjalny pracownik umie (a) coś wywnioskować z tego co już wie (np. odrzucić hipotezy o tym, co jest zepsute, albo zauważyć, że pewien sposób wyprodukowania zadanego systemu nie ma szansy mieć wymaganych własności), (b) tak wybierać pytania/następne kroki diagnostyki, żeby dostawać informacje użyteczne do (a). Niewielu kandydatów dobrze radziło sobie z (b); to że ktoś sobie radził było dla mnie bardzo istotnym plusem.

Show older
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.