Notices by Ignas Kiela (ignaloidas@not.acu.lt)

@dysfun@social.treehouse.systems I'm very opposed to the "just copy and paste these recipes" approach

Like, if it's just that simple, then it should be as simple as having a "standard library" of recipes and then having the user just define the recipe they want to use.

Of course it's not that simple, and we still want some kind of customization to be involved. Since we agree that the customization needs to be driven by a programming language, it's then a question of whether we value the ability for less familiar users to work on customizations well at the expense of how well the language meshes with the tool, or whether we value the language meshing with the tool well and having a nicer fit overall, at the cost of less familiar users needing to learn it to use the tool effectively.

My viewpoint is that for an ancillary tool like packaging is, it's a bit foolish to go for the second approach, because users won't have a lot of buy-in (it's just a "side" tool, it's not essential for their project). Learning another language is a barrier and a surprisingly difficult one (more from the psychological perspective than actual difficulty of learning a language in many cases). While with the first case some people will need to learn the language as well, the barrier is smaller because of perceived usefulness of a more popular language (I've seen a bunch of cases of "why would I learn [language X] if it's only useful for [tool Y]"). That way you're building a small number of [tool Y] experts, rather than building a whole bunch of people that know how to efficiently utilize [tool Y], and I think the second is much preferred.

"Decentralized" service fully censored by blocking a single server. Cool stuff, ain't it?

RE: https://mastodon.social/users/ooni/statuses/116453567258372626

@mnl@hachyderm.io @mjg59@nondeterministic.computer @david_chisnall@infosec.exchange @engideer@tech.lgbt the fact that something is random does not mean that it has a uniform distribution. "controlled randomness" is still randomness. Taking random points in a unit circle by taking two random numbers for distance and direction will not result in a uniform distribution, but it's still random.

like, do you even read what you're writing? I'm starting to understand why you don't trust the code you wrote

#ZeroTrust apparently means whitelisting MAC addresses

I will never stop being angry at security vendors simply redefining "Zero-Trust" to mean what they have always been doing instead of what it originally meant.

@georgia@netzsphaere.xyz @sun@shitposter.world the muscles needed are mostly on the back, what's on arms isn't doing much (though it's also not like modern bows are particularly hard to draw since they're designed more for accurate shooting rather than piercing through armor)

@quad@akko.quad.moe lmao

RenPy stays winning I guess

@sun@shitposter.world "automatically resets after 20 minutes"

techbros and their understanding of consent

Instagram reels sometimes bring some real gems of marketing

@mischievoustomato@tsundere.love @feld@friedcheese.us @quad@akko.quad.moe that's a classic tactic from patent trolls to set up a "patent pool" of stuff that supposedly something is using to scare companies into paying for it. AV1 has one for example, even tho it was explicitly designed to avoid patented shit

@mischievoustomato@tsundere.love @quad@akko.quad.moe this is not quite that level of stuff, I think some of these use opus underneath but there is stuff regarding how to transmit stuff between that impacts latency, error recovery and shit

@quad@akko.quad.moe bluetooth has like 4 different protocols for audio lmao, so probably that

@quad@akko.quad.moe there's a reason why I said "like" (tho didn't know there were 5 different aptX ones lol)

weird that even with the speakers being able to only run with a single protocol, it's still different, I'd guess somebody put some sort of filter in the android audio chain for phone speakers, yet it still gets applied for external audio as well

@feld@friedcheese.us @sun@shitposter.world ok thats bonkers and stupid

Both because it's on oracle, and because SQL is kinda ass for such code volume

@azonenberg@ioc.exchange i don't think ypu can do it (well?) with Vulkan but it is possible to do GPU-side allocation. Not sure if worth tho.

@sun@shitposter.world @feld@friedcheese.us how often you're transitioning between DB vendors anyways?

(also, since writing them is litteraly part of my job, the best use cases I'd say are either small ones for ensuring nore complex consistency terms of the database, or large ones for bulk data transformation and saving.

@sun@shitposter.world @feld@friedcheese.us ok, I understand it with oracle because of it's toxic licensing

But like, I've rewritten a whole bunch of stored procedures - e.g. from postgres to mssql and then back to postgres after some changes while it lived in mssql. Was fine.

And yes, if you put all of it in stored procedures, you're gonna have a bad time migrating (guess why oracle wants you to use them so much), but as I said, you should have either small ones for consistency, or large ones for large transformations/analytics. The first case is easy to rewrite because small, the second case is a pain either way because the alternative is just having all of the sql in some script or whatever and you're gonna have to rewrite it either way.

@sun@shitposter.world depends on the human not having 100% trust for the LLM's results tho, like seemingly a whole bunch of curl's bug bounty reports were humans that just had absolute trust in LLM's results and even tho the talking didn't seem LLM, the arguments and reports did seem to fit a LLM

@niconiconi@mk.absturztau.be the real question is when will we stop discovering new adhesive tape radiation wavelengths

@whitequark@mastodon.social this implies that clang is a mammal and can get pregnant