debugnik

I'd rather talk about whether the protests were "genuine" rather than "legitimate". Riots usually escalate from prior peaceful protests, in which protestors claim specific concerns and political factions get involved. None of that seems to have happened here, it was zero to teens rioting.

Didn't we get a submission today about Qwen3-Coder producing more bugs when hinted on the user having different ideology than the Chinese government?

Ok I see it's flagged now, probably fake, but the fact that it's possible does make me feel uneasy about both Chinese and US models.

Only so-called "ref" (and "out") variables/returns/fields can point to the inside of an object. These follow some borrow-checking rules so they can also point into stack variables, thus only ref-like types like Span<T> can hold ref and ref-like fields, and these are always rooted in the stack. But refs into the inside of a class type can escape the function, because they can keep the object alive.

The company, Calif, has it with the music, effects and Ui Shigure animation on their own website and youtube channel: https://calif.io/php2shell/#php-5-4-0-rce

I know a professional board-game designer who's using AI to make web prototypes now. He doesn't understand JS yet he still manages to model the gameplay he personally comes up with. I'm coming to terms with the fact that slop code can satisfy non-slop functional requirements, and the latter is what players will identify as non-slop.

But I don't think non-slop art is separable from the process to make it. I do however think gen-AI is blurring the qualities that distinguish non-slop by proxy, and we'll end up trying to find meaning in non-existent design/art decisions. I've already stopped trying to discover new digital artists for this reason. One day I might drop out of gaming beyond casual play with friends at this rate.

I suspect they're using some new cheap multi-modal model. A couple of times now I've seen the generated subtitles fill muffled speech with text about something unrelated happening on screen.

I'd say gen-AI art assets are definitely the deep end, and DLSS-Yassify is right behind it. I'd put code behind them because there's a second gradient within it from Copilot-on-a-short-lease to vibe coding.

I used to assume that low effort indies would be the major force for the game equivalent of gen-AI tiktok fruit slop, but when I saw DLSS 5 I realised it might come for AAA first.

> If you combine partial parsers, you end up with a partial parser.

If by partial we mean the parser can fail, not necessarily. optional(elem) and repeat(elem) are total if elem doesn't consume input on error, at least for a reasonable semantics of parser combinators (although we usually consider it a programmer error if elem matches the empty string in repeat(elem)).

Not in Spain, we only use electronic means for preliminary aggregate results.

Each voting station gets a paper voter list. When one votes, their name is struck off this list and added to a new one by different people. Then ballots are tallied one by one in front of observers, and the count and both lists are checked to match.

Finally, the preliminary tally is sent, but physical evidence is brought to a nearby court by the station members for custody, plus a copy of some documents sent by mail, so the electoral boards can aggregate the results during public sessions.

(I too use Firefox, even on Android.)

I just don't like this idea that downstream maintainers need to be saved from themselves by limiting the options upstream provides them.

Downstream are not end users, which I agree should only get tested builds. They're professionals too, and they can be trusted to pick the configurations they want to build, evaluate how much more testing they need, and cut out configurations from their support matrix if they don't have the resources to support them.

Also, this idea that a professional project wouldn't offer untested combinations of settings to other professionals is why people follow up asking about Linux. A quick search tells me it's got ~19,000 Kconfig options[1].

[1]: https://dl.acm.org/doi/fullHtml/10.1145/3646548.3676548

I can tell you what Gentoo shouldn't do: blame upstream for having assigned themselves the mission to support and test 2^N combinations of options for which they don't get to curate N.

They're free to limit which options they offer, they're free to limit which combinations they support, and they're free to fork or drop the software they offer if they insist on keeping the 2^N purity (which they eventually did). But that's Gentoo's mission, not upstream's. Upstream is just giving them software for free on an as-is basis.

The data is mostly teens and young adults pushing the bots into softcore smut as far as the censor classifier allowed them. Almost a liability rather than an asset.

It's a roleplay app, an AI version of the old RP forums, or RP private chats. Anyone sane and slightly mature understands they're not being encouraged to do anything IRL. The main problem I see here is a lack of parenting.

They were asking for a reference/source, not to be treated like an idiot.

"He" isn't even covering it, he eventually automated himself out of the channel: He AI-cloned his voice and had it read AI-generated scripts. I assume he still does.

Wii U and Switch aren't x86 at all. All of GameCube, Wii and Wii U used a PowerPC CPU and an ArtX/ATI/AMD GPU. And Switch has moved to Nvidia ARM chips; they had been using ARM for portables since GBA.

We recently got an article claiming the front and back of the brain grow from two different progenitor cells, which some scientific media reported as it being two organs, whereas others claim that's an exaggeration.

I was assuming they meant for the heap. Not even executable memory is locked though, so if you mlock too much memory the kernel will page out your executable instead. Something has to give, so it's better to schedule it.

On a lower level, the OCaml and cpython GCs use a prefetch buffer during marking to schedule around the cache.

Aren't madvise and mincore the APIs you want?

Thanks, that makes sense. I was vaguely aware that SHA-256 halved the internal state from SHA-512 but not that it halved the block size, I haven't looked much into it. Although it makes sense in retrospect that they would simply use 32-bit variables to keep the same structure.

C has many more ways to trigger undefined behaviour than memory access. If C had unsafe blocks they'd restrict most forms of signed integer arithmetic and shifting, for a start.

How could SHA-512 be faster? It does more rounds of the exact same operations as SHA-256 with a bigger state. Although if it really were faster, SHA-512/256 gives you a truncated version.

Good question, why did Russia rename the Shahed-131 to Geran-1? Those are allegedly Iranian made, and Geran-2 is allegedly a slightly modified Shahed-136. Of course, as they keep modifying them they slowly become their own thing, but there's a clear tech ancestry.

I don't think of the stdint types as being less native than the keyword integer types. C# for example makes keyword types aliases to System.* types, not the other way around.

long can't be smaller than int, so a 64-bit int leaves short as the only type between char and long long, and you need to pick whether it's 16 bits or 32.

Then again, we'd still have stdint around. And ultimately this doesn't matter because most code isn't portable anyway.

I didn't intend to argue either way, but I feel that adding "and search" to the argument made above doesn't really weaken it. The full claim is that regular users weren't really interested in using power-user apps, so they won't be interested with a chat-bot UI either.

I do personally think that we'll see regular users interested in new, different workflows than the automation ones we associate with power users, and the image generation use case you mention is a good example. Chat-bots do seem to be owning those rather than entirely new apps.

The use cases described above aren't really multiple applications though. They're just using a search method to get the right fact or factoid. I consider that the scope of a single "app".

Most of those websites have been complements for Google Search rather than standalone services whether they like it or not, and if chat bots replace Google Search, they'll replace the entire ecosystem.

Sounds like it's replaced Google Search for her, not much else.

I've been seeing "AI;DR" used sparingly for months, perhaps years now.

> I wouldn’t chalk any of this up to malicious intent. I’m sure they are swamped with such requests.

Stalling in the hope that reporters won't escalate, instead of allocating a tiny bit of their billions in profit to hiring for this, is malicious in my book.