@[email protected] тоже прикол с whisper-ом)
каким-то образом единственное нормальное объяснение что я быстро нашла - на ОТВЕТАХ МЕЙЛРУ
https://otvet.mail.ru/question/237076673
@[email protected] @[email protected] my chemistry teacher used that phrase a few times, and i always found it an incredibly unsettling thing to say to a group of children :neocat_woozy:
this stupid spyware decided to stop working.
it just ignored any requests from localtuya.
after another hour of debugging it turned out it hadn't heard from china for too long apparently. re-enabling access to china fixed it :haggard:
fucking chinese spyware.
i should finally get zigbee bulbs...
@[email protected] i honestly feel like people doing web standards don't care about performance like *at all*. and like, it *can* be justified, people don't usually do heavy stuff on the web.
but the issue here is that js is definitely not only "the web". its used for *everything* these days. yet without node-compatible apis, your runtime is useless for backend (yes, im talking about deno).
and it's such a pain trying to make your library support all the different runtimes, while also utilizing runtime-native apis for maximum performance.
i can't just use `TextEncoder` because of performance penalty.
yet i also want to support web.
and i definitely don't want to use a 28kb large `Buffer` polyfill for browser. :neocat_woozy:
another example - webcrypto is **very** slow. on node they are 5-10 times slower than `require('crypto')` apis. despite both being openssl underneath, lol.
wow. `TextEncoder` api in node is about 2 times slower than `Buffer` :neocat_woozy:
[apparently](https://github.com/nodejs/node/issues/50485#issuecomment-1786915761) `Buffer`s use a buffer pool, and `TextEncoder` creates a new `ArrayBuffer` each time, which is slower.
web apis suck so much smh
@[email protected] when you're ill to the point where its difficult for you to even go eat something or take a shit - going outside is the last thing i would consider doing 😣
@[email protected] it's definitely less of a problem e.g. in rust simply because rust is not a language one can pick up in a few evenings without any prior experience.
though i do agree this is a people problem. let's be real, 90% of people are fucking stupid. having a slightly higher entry barrier at least keeps the stupid people away.
@[email protected] nah, that's not what i mean. for the lack of proper stdlib we have lodash. that's honestly the okay-ish part of npm. while there are people like sindresorhus who i believe are poisoning the community, such utility libraries are mostly okay.
what i mean is that there are too many people who learn js on some shitty online courses and write absolutely useless unmaintainable untested spaghetti code. and then publish it to npm because why the fuck wouldn't they, the entry barrier is also non-existent basically. even if it makes zero sense to publish this to npm (like what the fuck is [this](https://www.npmjs.com/package/desktop)).
and then there are the god complex individuals who push fucking obfuscated blobs to npm. and what bugs me the most is there are actual human beings out there who use them (i won't point fingers, but this thing in particular has 500+ weekly downloads smh).
yet once you need something more niche you're mostly out of luck, because most of the time all existing libraries are either incomplete, broken in some way, outdated, unmaintained or just unusable. and ive seen that pattern too many times sadly :c
@[email protected] e2ee on its own means nothing.
you must at the very least verify that the encryption key is the same. ideally in person.
and even then you will have to trust that a) the app correctly displays the encryption key, b) there are no backdoors in the app itself. and somehow make sure nobody can steal your local database of messages.
e2ee is not a panacea.
@[email protected] yeah that's probably one of the reasons i am here :з
mh~
$[blur another being trying to deal with my social anxiety (which honestly kinda goes well)]
@[email protected] yeah, ik, its just that...
idk, i feel like some part of me just refuses to be silly. like its bad or something..
i guess cringe culture has gotten to me 🥲
ive been working here for more than a year already and still get extreme anxiety every time i have to interact with a colleague outside of my immediate circle
why are people so scary,,,,,,,,,,,
<small>i should really stop venting here smh</small>
@[email protected] tbf both are bad
because nobody is stopping you from making a package `VeryGay` and using namespace `Gay.*`.
and similarly in java. anyone can use any namespace, which sucksssss. too much mental overhead. and also good luck with conflicting transitive dependencies.
npm and cargo are probably the only sane package managers, since they force you to import from the same name as the package name. you do `npm install very-gay` and can be sure that `import 'very-gay'` will do exactly what you expect
about the domains in java - afaik, some primary java package registry requires you to prove ownership of the domain before you can publish a package.
@[email protected] so true
i hate that it increases bundle size for everyone
and loading that part of code on-demand in some cases introduces insanely unstable crutches
i needed to run a wasm-based lib in a service worker only in safari, and i kinda gave up on loading it dynamically
service workers can't use `import()` nor `importScripts`, and when i tried to make a crutch that would communicate with the page for that - it was so flaky i just had to give up and import it directly
i could probably generate a separate sw bundle for safari, but i didn't feel like hacking around with vite either :neocat_woozy:
bun is so cool
i have no idea how is it compatible with native node addons, but it is and its 🤯
i changed a total of 0 lines of code in my code for all 1k+ tests to pass
so cool...
@[email protected] just make an input and put it as a checkout ref? don't think there ever was such a button, can't find these old docs either
smth like this: https://stackoverflow.com/a/75296382