@treuss I'm afraid they exist whether I'm into them or not. :)

@mrrmot @1dalm I'm not telling anyone what to do and what not to do, but *I* do not intend to check any thongs.

@vbabka @tost Nothing interesting, just someone linking to a random driver bug they are mad at. Unfortunately, HN has a bajilion bots that all want to load a preview of the linked page, and that nukes bugzilla dead.

@tost all right, let me try this. Thanks!

@tost ORLY. Okay, do you mean pleroma.web.preload or some other setting?

@tost We're doing okay for CPU -- it's really the DB load that has the most impact on site availability. Now, the DB is a pretty small instance, so it does make sense to throw some more CPUs at it to begin with.

@tost Oh, it was for bugzilla, not for Akkoma.

Thanks for the DDoS, HN. I didn't have any plans for the afternoon anyway.

@Hyolobrika yes, but instance-wide only a handful of clearly offensive trolls.

@Hyolobrika yes

I've blocked more accounts in the past day than in the entire past 4 months.

@drq Yes, I've heard that before. Hopefully, we'll find a way to deal with that.

@alex I'm quite happy with Akkoma's UX.

@adiz Sorry, I meant Typescript, not PHP. I was thinking GnuSocial.

@adiz PHP is less fun than Erlang. :)

I picked Akkoma because it looked more fun. No other reason.

Heh, I can tell when Linus posts by the amount of 500 errors I start seeing on the social system. Need to start optimizing our setup.

The erlang stack itself is actually doing great, but the database load spikes like crazy. It's actually a known problem of pleroma and derivatives -- the db backend could use a lot of loving, especially when it comes to optimizing queries.

Here's my _personal_ view on the Baikal Elektroniks situation. This is not any kind of official statement by LF.

First of all, everyone must understand that Baikal Elektroniks is a company that produces equipment for pretty much a single client -- the Russian state. You can nominally buy a computer with a BE chip as a private citizen, but in reality you'd never do so because a) it's almost impossible to get, b) you'd buy a much slower chip and pay 4-5 times more than you would for any other chip available on the market. So, it's accurate to say that BE produces equipment pretty much exclusively for the Russian military and its state run businesses (who are mandated to buy BE equipment by law).

Second of all, and most importantly -- getting your patches accepted into mainline means receiving a lot of very expensive labour and computing resources gratis: you not only get free code reviews from maintainers, but you also benefit from a bunch of behind-the-scenes CI infrastructure that runs checks on your code -- both at the patch stage, and later as part of regular integration/CI/fuzzing runs. Any treewide changes, such as security improvements by efforts like KSPP, will also be automatically applied to any in-kernel drivers and architectures.

So, in reality, accepting code for any hardware into the Linux kernel means helping to test, maintain, and debug that code for years to come. The resources for that are pooled from many device manufacturers with the understanding that these efforts will be part of the tide that "lifts all boats," including their own. However, in the case of Baikal Elektroniks the situation becomes tricky. Yes, Linux is free software (free as in libre), but maintainers and CI infrastructure require funding. BE is placed under strict sanctions in many countries due to its direct affiliation with the Russian military, so companies funding CI and maintainer efforts have to consider if their money is directly benefiting a sanctioned company (and, indirectly, the Russian military).

So, it may be true that the rising tide lifts all boats, but if that boat is a Russian military warship, you have to decide what kind of message you send them.

@dirkhh not to downplay Mastodon's accomplishments in any way, but the backend software on social.k.o is actually Akkoma, not Mastodon.

@cliffwade @iamstu Okay, but the question was what kind of link -- to what site -- would be a better proof than the account actually being on kernel.org?