Notices by wohali 💯 (wohali@octodon.social)

@luka yes! easily.

#OH #IRC

a> i asked my boss if i could wear an 'impeach trump' pin/button to work. he said no and laughed

b> put a slash between them and claim "these are my pronouns" et voila

@Seabird big mood

@ark the only things i heard were praising ryan reynolds' voice acting, which, sure, whatever.

@ark i watched half of that movie without audio on a plane recently and thought THE EXACT SAME THING

@taweret this has been going on in collectibles for decades now

and yes, fuck capitalism

@libc

in C it was always bitmasks with `& 0xffff000` or `& 0b0011111000`, you could make a convenience function that would take a low and high value and build the bitmask for ANDing I guess, but usually I just inlined it

in python i'd use bitstring or C-like approaches

C++ has std::bitset but you're still doing bitmasking like C so I'm not sure it does much for you other than type safety

@libc Here's a tutorial on how it's done in the free-and-open-source KiCAD.

https://www.baldengineer.com/kicad-bus-labels-and-global-labels.html

@elmiko Everything I said before + more that I charge for consulting for ;)

@libc here's an example from Eagle, where a double-dot ellipsis on the netlist is used. Only change is the order is [low..high]

Context matters but honestly if you say BUS[15:8] or BUS[15..8] everyone will know what you mean.

@libc usually net[7:0] for Verilog notation or net[7 downto 0] in VHDL.

Verilog is the more common syntax and you'll see this type of notation on schematics more often in my experience.

@elmiko @djsundog @KemoNine what I do for packaging Apache CouchDB is that I build packages (.deb, .rpm, etc. + .tar.gz) that work on as many platforms as possible.

Our Docker container is just a Linux container that installs the .deb package into a container and sets a few network settings.

It works. But it's measurably slower than running the same on bare metal, even for a naive untuned setup. For a tuned setup it's significantly slower.

@elmiko @djsundog There was a long Intel whitepaper that went into detail on this at one point, but it was taken down and replaced by something that's more of a sales pitch for their specific workload management stuff...which is less general purpose and requires coding to a specific standard

kinda misses the point

@elmiko @djsundog No, because we're talking about pinning to specific cores on each machine, in conjunction with the NICs on those machines, which must also be pinned accordingly - not just "land this on a machine that has the right GPU."

@elmiko @djsundog anything that expects very low latency between network traffic and disk activity (i.e. serving database requests in <5ms, preferably <1ms) you have to worry about pinning interrupts and processes to specific cores. Beyond this you have to do careful planning of how bw is allocated between local storage (again, latency, not bandwidth, so SANs are unacceptable) and processes runnign on machines to avoid the "noisy neighbour" problems.

Way beyond what k8s/swarm can do today.

@nev can breathe easier after seeing this earlier today.

still, shit is f'd out there. especially in hamilton.

@lyliawisteria I haven't. But I can stand against it.

@interneteh I'd play it