Katy Rubin gives an energetic defense for #democracy in this podcast:

accidentalgods.life/peoples-ch…

Our current “democracy” is not fit for the challenges of preserving complex life on Earth—it functions more like a kleptocracy. Ordinary people need to reclaim governance by building what they call a “House of the People,” where wisdom—not wealth or power alone—drives decision-making.

It's a hopeful discussion of what is possible, and indeed what is actually being done.

Maybe if we stop behaving like we're POST-COVID-19, eh.

PS Sick days could be rising again due to people’s increase in exposure...

Number of sick days taken by public servants growing post-COVID: globalnews.ca/news/11357829/si… #COVIDISNOTOVER #KEEPVACCINATING #WEARAMASK #cdnpoli #polcan #polQC #QCpoli #polMTL #MTLpoli

We updated our #akkoma integration page with an easy way to link your account to a Prosody #xmpp server: joinjabber.org/tutorials/integ…

Thanks to @nigel for testing it.

@akkoma maybe something to add to the official docu as well?

Peter Vágner reshared this.

in reply to Daniel Gultsch

I assume as long as SASL-SCRAM-plain is the only way to achive auth integration with other system, there is really no way around that. Channel Binding is a nice feature, but personally I find it much lower priority than auth integration.

Maybe you could look into supporting Oauth2/OIDC login flows in Conversations? At least Prosody seems to have good support for this now, and I think this might be the only realistic way to have both Channel Binding and auth integration.

in reply to Kris

@kris As far as I’m aware the oauth support in @prosodyim is for authenticating other apps against existing prosody users. Meaning the user database of Prosody would be the source of truth which would allow Conversations to use channel binding. So yes I agree that this would be the better approach to integrations. But I don’t think Conversations is much involved here.

The federal judge let Google off the hook in the antitrust case that the company supposedly lost. He said no to any serious remedy. And he indirectly killed Mozilla (Firefox and Thunderbird).

A good day for Google, and a terrible day for what's left of the open web.

arstechnica.com/gadgets/2025/0…

After an amazing long run of stability it has been decided that I am back on the job market and available to work for a new employer. I prefer to remain in the #accessibility space, but I guess beggers can't be choosers and all that jazz. I'm doing my own search but if anyone out here is on a team needing help or has any job leads, I would really appreciate making the connection. #CPWA #WCAG #Salesforce
in reply to Darrell Hilliker 👨‍🦯♾️📡

The NFB recently advertised two open positions in accessibility advocacy that were circulated here on the Fediverse. I assume you would need authorization to work in the United States for those roles, so they may or may not apply to you. I looked at them myself, but I'm not looking for advocacy work - rather, I wish to be in more research and policy-oriented positions that ideally also emphasize contributing to the scholarly literature (e.g., academic or policy organizations).

For a service that depends directly on a Postgres database they've shown:

- they don't know how to properly manage storage
- they still never turned on pg_checksum
- they have no idea how to run a reliable production Postgres cluster

These are unserious people trying to run a serious project and it should make you very concerned about how professionally they do all their work
RT: mastodon.matrix.org/users/matr…


Sorry, but it's bad news: we haven't been able to restore the DB primary filesystem to a state we're confident in running as a primary (especially given our experiences with slow-burning postgres db corruption). So we're having to do a full 55TB DB snapshot restore from last night, which will take >10h to recover the data, and then >4h to actually restore, and then >3h to catch up on missing traffic. Huge apologies for the outage. Again, folks using their own homeservers are not impacted.

Peter Vágner reshared this.

in reply to feld

Oh a mysterious "slow burn" of Postgres corruption? Where is the engagement on the Postgres mailing lists? I haven't seen a single thread about this issue on the pgsql-general or pgsql-hackers lists.

It's either a hardware storage bug, a raid implementation bug, a kernel bug, or their Postgres/filesystem tuning is trading data reliability for performance. But they're not sharing anything of value.

Postgres doesn't just corrupt itself. We have several DBs > 100TB at $work. Many people have significantly larger databases...

I kinda doubt their recovery times too. They will probably forget that they need to disable indexes to make the restore have a reasonable speed. And pg_restore is single threaded per table. 1.5TB can take 1.5 days.

blog.peerdb.io/how-can-we-make…

I think they're fucked. I wonder if they will be able to recover without it taking months, literally. They haven't indicated they're using anything but vanilla Postgres.

This could be the end of the matrix.org homeserver.

This entry was edited (1 week ago)

Peter Vágner reshared this.

in reply to Blurry Moon

@sun Oh I don't disagree, but Oracle has had billions poured into it so they can make that possible. Postgres is nearly as good as Oracle in almost all use cases, but these types of maintenance operations have not yet been engineered for performance.

The companies doing the Postgres forks have been the ones innovating here and putting their time and expertise into making sure they solve their customers' needs. And often those improvements get merged upstream. But as far as R&D goes it's still a drop in the bucket compared to Oracle 🫠

in reply to Dr. Cat

@j @sun At that point you basically consult them for every patch / upgrade so they can keep your changes working. This is sometimes a major issue keeping companies from upgrading to the next major release as their query planner etc will change and you could lose the performance you had or new problems arise that need different custom patches to keep your workload performing as expected.
in reply to Dr. Cat

@j if you aren't a big company the few benefits aren't worth it, in fact I would say it has little benefit unless you buy their most expensive horizontally scaleable option which is meant for busineses where the data size is so massive it should be nosql but you're architecturally locked into rdbms. very time I've mentioned it on here people say "you're doing it wrong" well I have to explain that a lot of corporate customers are just plain locked into somethigng that got built in the 1990s and it would take a hundred million dollars and shitloads of uinacceptable risk to rewrite. for those customers there is a big fat oracle database and you will pay a LOT for it.
in reply to Dr. Cat

@j for years and years people went with oracle because it was the only ANSI SQL compliant database, everybody else either didn't have x feature or it was a proprietary extension. but this hasn't been true for years, Postgres is compliant.

oracle also spends a gazillion dollars convincing your company to put everything into oracle though, so they have really stupid bad shit you should never do, but on the surface you think "I'm already paying them so I'll integrate that too". it's pretty transparent that they're taking advantage of know-nothing managers to trap companies into never being able to leave.

in reply to Blurry Moon

@sun @j We had a scheduled overnight outage in 2007 to upgrade Oracle 9i to 10g. It was an 8 hour outage and the process to backup then apply the patches took 7 hours.

We couldn't afford more Sun servers. A restore from backup was also 8 hours. We practiced it several times because even doing one thing out of order breaks the database.

It was all or nothing (and probably losing our jobs).

It worked. I was never so scared though

in reply to Blurry Moon

@sun @j You can and I have done so without issues, it's just not a simple install, configure Postgres and let it run thing. You have to change record sizes to avoid fragmentation, if you are on fast SSDs disable ZIL on your DB dataset and hope that Postgres will ensure data integrity with fsync and pg_wal, or move it to a special ZIL SLOG on fast SSDs. And those are the absolute basics of what you have to do to make it somewhat work.
in reply to Blurry Moon

@sun @phnt here's this too
vadosware.io/post/everything-i…

Setting record size to 8k is faster than 16k but only for a little bit because it gets super fragmented. Setting to 16k fixes the fragmentation and provides better compression ratios since compression happens to each record block. Setting to 32 or higher could be interesting and help compression even more. You won't see improvements beyond the default 128k on like 95% of drives and it could even hurt performance. That being said 1M+ record sizes may be useful in conjunction with zstd-4 for long-term archival of compressible data like database backups. All of this can be changed whenever so it's not that big of a deal. Block size you're stuck with forever so make sure you set the correct block size.

Most of the data the database actually cares about at any time will live in the arc (ram cache) and if you use compression it's compressed in the ARC so you get even better cache hits.

For compression I used lz4. Zstd (even compression level 1) was too much latency. Lz4 is really great and shaved off about 45% of data needing to be written to disk. That was the main reason I switched to zfs. It was the only practical filesystem for postgres that supports disk compression.

It makes postgres upgrades super fast and easy. Just take a snapshot, hard link the database files, fire up the new postgres version and it should work but if it starts fugging the database then you can just easily restore the snapshot.

I came for the compression and ending up loving it because not only is it the best filesystem but it's the best disk management system too. You can even just create raw volumes and format them however you want. You can have ext4 on zfs, you can have NTFS on zfs, you could even put zfs on top of zfs if you really wanted to.

Zfs is also the only way to have a compressed swap partition

in reply to Dr. Cat

> All of this can be changed whenever so it's not that big of a deal.

when you make these changes to ZFS filesystems it does not change the existing data. That problem is left to you to solve -- traditionally by restoring all the data from backup.

However, a new tool is coming called "zfs rewrite" that will let you atomically rewrite underlying blocks so the data gets the new storage settings applied to the filesystem.

openzfs.github.io/openzfs-docs…

edit: this would also be useful for re-balancing your zpool if you add new zvols or something

This entry was edited (1 week ago)

> “(T)he Court’s role is to respect the choices that the people have agreed upon, not to tell them what they should agree to,” Barrett writes in “Listening to the Law,” set to be published on September 9.

So if the Court believed the majority of the country thinks slavery should be legal -- whether or not that's actually what the country really believes -- they'd rule it's legal again? Like, what the fuck is she on about? Their job is to interpret the Constitution not rule in favor of what they believe the majority wants

#InEarMonitor #IEM #audiophile
Here's a mind-blowing test track for headphones/earphones/earbuds. It's an incredible experience. You have a giant speaker rotating around your head and Billie stomping around you in a circle clicking her fingers. It will not work even on a good surround sound system.

8D AUDIO PENTATONIX / Billie Eilish - Ilomilo (USE HEADPHONES) whatsapp audio
youtu.be/-tRk9N8teLU?si=tPn30C…

uspol

Sensitive content

This entry was edited (1 week ago)

The US government abandoned antitrust. Today, companies facing antitrust can pay key Trumpland figures a million bucks, and they'll make a discreet visit to the 5th floor of the DoJ building, have a shufty around the Antitrust Division and the whole thing will just...go away:

prospect.org/power/2025-08-19-…

--

If you'd like an essay-formatted version of this thread to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:

pluralistic.net/2025/09/02/act…

1/

Will anyone be updating the NVDA AudioScreen add-on for renewed compatibility with more recent versions of the NVDA screen reader from NV Access? E.g. keyboard shortcuts appear reassigned github.com/nvaccess/audioScree… @NVAccess #blind #a11y #accessibility

Once up-to-date again, I'll gladly add a link to the NVDA AudioScreen add-on in the white paper on brain implants for the blind versus visual-to-auditory sensory substitution artificialvision.com/neuralink… Global accessibility matters.

This entry was edited (1 week ago)

I'm failing to log in to one of my gotosocial instances.

The account credentials in my password manager aren't working.

I tried resetting the password from the command line, and that succeeds, but I still can't log in.

I'm wondering if I'm using the wrong email address, but I can't see a clean way to find the email address associated with a GTS account from their cli tool, and glancing at the db schema also proved fruitless.

Kerncentrales zijn belangrijk voor hernieuwbare energie!

De locaties zijn namelijk voorzien van een superieure stroomaansluiting. Nadat de reactoren gesloten zijn, staat er in Duitsland een kleine 3GWh #batterijopslag gepland
Dat komt qua capaciteit in de buurt van de "Hydro Pumped Storage", een belangrijk fenomeen om de hernieuwbare dagcyclus te stabiliseren. (Na de dagelijkse portie zonnestroom is er nog een vraagpiek in de vraag rond etenstijd)

De zonnepaus spreekt:
social.anoxinon.de/@solarpapst…

Inak, ze preco spominam nasilie:

mastodon.social/@phanecak/1151…

Pan terazky minister vraj o.i. povedal cca:

- "Štát chce znevýhodnených, nízkokvalifikovaných alebo dlhodobo nezamestnaných zamestnať prostredníctvom sociálnych podnikov."
- "Na prvých projektoch sa už vraj Tomáš dohodol s ministrami životného prostredia, pôdohospodárstva a dopravy. Pôjde napríklad o sezónne práce na farmách, čistenie lesov a riek alebo kosenie okolo ciest a staníc."

1/2

in reply to SuspiciousDuck

@SuspiciousDuck Ved to! Ma to mat dva ucely:

1. Nezamestnani maju mat pracu (=prostriedok) a vdaka nej prijem, z ktoreho mozu primerane fungovat (=ucel)
2. My ostatni mame mat vdaka tomu lepsie a prijemnejsie miesto na zivot.

Ale ked to robime takto, tak:

1. Je to drahe
2. Niekto bohaty sa na tom nabali este viac (pointa existencie #SmerHlasMafia)
3. Nezamestnani zivoria a su nestastni
4. My ostatni to platime, ale tiez sme z vysledku nestastni
5. bonus: poslapana ľudskosť

> while overlooking that other instances of human language use also miss that feature.

This to me always seems like how many people strongly believe that only humans are self-aware and can experience a full range of emotions (love, grief, etc) even though we have empirical evidence of many different species exhibiting those behaviors.
RT: fediscience.org/users/UlrikeHa…

As with the demise of windows10, various meta happenings around Whatsapp open possibilities to welcome refugees. Instead of blaming them for not having had whatever insights earlier, there are often just very practical questions where a helping hand does wonders. Offering help instead of trying to convince.

Currently many companies are giving win11-incompatible laptops away, often powerful systems that deserve upgrading to Linux, and then donating then. It also helps to counter hardware waste.

does anyone know the keyboard command to show/hide the video window in winamp? when my VLC isn't streaming to Kalvins tv, I use that for videos, but when it is, I need to use winamps ability to play videos. normally if I open an mp4 file, the video window normally automatically opens, but lately it isn't for some strange reason. either that or it doesn't work with meta glasses videos for some strange reason.

There are little things I hate more than a "free" service that becomes paid when you go over an API limit and wants your credit card, even if you're only planning to use the free plan.

No, OpenWeatherMap, I don't care that you can set a request limit in your UI. I'm one missed "The free request limit changed" or software bug away from a crying wallet :/

in reply to modulux

@sukiletxe @lehnautora

Yo tengo la sospecha de que Bello usó esa terminología para enseñar en América la Consecutio Temporum (concordancia de tiempos) al uso de España. En las Américas, los tiempos verbales se usan de forma diferente a España, y puede que el problema de mal uso del condicional en prótasis condicionales fuuera menor que otros problemas surgidos de la acomodación entre tiempos. En cualquier caso, en aquella época lo llamaban "potencial".

(Hay que recordar que el condicional no solo se usa en condicionales, sino también en concesivas —aunque mis profes me tiraron de las orejas cuando les dije que las concesivas son condicionales falsadas—, en peticiones, en expresiones "potenciales" que indican suposición referida al pasado —si se refiere al presente, va en futuro—, y en muchas otras ocasiones).

Uff, qué rollo he metido.

This entry was edited (1 week ago)
in reply to Microblog Castellano

Claro, es verdad que el condicional tiene varios usos, como el propio presente.

Dicho esto, lo de las concesivas creo que también se sale del antepos, ya que la modalidad típica es utilizar el condicional frente a un presente: por mucho que llueva no usaría paraguas.

Por cierto, hoy en día ¿sigue habiendo oraciones bipolares, o se consideran subordinadas?