@lain I mean, I did explain exactly what I needed it to be able to do, which file to look in to add the functionality, and exactly what I wanted the configuration file syntax to look like and where the docs/man page was that needed to be updated.
and it did it. it just worked. I've read through the code about 5 times now and I don't know all the quirks of OpenSSL obviously, but there's nothing strange or scary going on
@lain i mean i guess if you open a new empty folder and say "Qwen, build me an MVP for an ecommerce website" and walk away you're going to end up with a giant stinking pile of shit
that seems to be how the average hater assumes the tool is used
This release is targeting the Desktop UI and uses SmartDock along with many of the configs that Google included in the AOSP 16 source, along with a few FOSS app stores and MicroG.
BassRPI 16 QPR1 (Bass ARM) Desktop PC - FOSS - RPI5, RPI500, RPI500+
Android 16 QPR1 for your RPI device that comes in a number of different flavors (Tablet UI, Desktop UI, Kiosk UI, etc)
This proj...
Re last: That's why I'm constantly preaching about it. All of you, "let's ditch #Windows for #Linux" enthusiasts, automatically exclude blind and low-vision people which are already vulnerable on the job and education market. #Accessibility
Yet another reason I'm glad I don't use Linux. It's great software, but accessibility is so unpredictable and you don't have access to apps that are Windows only. Nothing against those who choose the Linux route, but it's just not for me.
I understand why, but the abysmal state of accessibility on Wayland means that a part of the open source community that depends - and I don't mean "uses for work" I mean "depends on to be able to use the machine at all" - are getting hung out to dry.
Somehow the free software community chose enterprise-infosec management-by-checklist over human centered usability and I'm worried we're going to be paying for that choice for a generation.
Attached: 1 image
#Plasma is going all #Wayland
After nearly three decades of KDE desktop environments running on #X11, the future KDE Plasma 6.8 release (due early 2027) will be Wayland-exclusive.
Seit ich neulich mit Rückenschmerzen aufgewacht bin, gucke ich mir den Aufbau der Hotelbetten genauer an und stelle fest: fast überall soll ich auf einem Brett schlafen :( Das finde ich für Camping, Hostel und Jugendherberge ok, aber in Hotels, die auch Hotelpreise kosten...?! Freue mich wenn jmd. eine Empfehlung für eine (bezahlbare) Kette mit anständigen Betten hat.
I'm trying claude.ai for the first time. I already miss ChatGPT's interface. ChatGPT uses level 6 headings for each side of the conversation, so I can quickly jump to the latest response or review what I wrote and then skip down to the response. It tells me when it's generating, and when it's done. Claude has neither feature. But it's supposed to be better for coding, so here I am.
Tonight join @Liz, for Full Circle from 7 P.M. eastern (US time). Their will not be a showcase this week, as Liz has the Full Circle Pop Up Comedy Club. In the third hour, she will have this weeks episode of Chickenman. So tune in tonight, for three hours of comedy. HKCRadio.com
SFC just launched our biggest match challenge! Can we raise $211,927 before January 15? (Those last days overlap with the Vizio trial!) I'm excited to find out.
Software freedom is critical to many of today’s most pressing social issues, but it’s only effective when FOSS is for everyone. Support Conservancy today to help make that happen!
#GrapheneOS for #Pixel10 is finally here! The installation process was straightforward, but the most impressive aspect is how it restores (almost) all applications, along with their data. This is done from a backup, so you can do it even if your previous phone is lost. All you need is WebDAV storage. If only all Androids had this!
well. New Androids work exactly this way. You just login into your account and it can install all aps you had in older device. Samsung has it, if not other vendors.
if the applications store settings in Google drive, them it restores it as well. Depends on application. If no online backups are made then you won't have data to restore. I think android has it good in recent versions. Much easier to switch to the new phone, compared to 4 years back
@monk @pemensik yeah, I think saving app data to Google Drive isn't really acceptable for someone who uses GrapheneOS because then why to use GrapheneOS in the first place? So I'm glad GOS provides a solution to that.
The migration story on Android may be better than 5-10 years ago, but it is still light years behind iOS, or even desktop Linux where the process is much less manual than on Android. And it's deliberate. Google could have come up with something similar, but they chose not to because their interest is to sell cloud services.
Are you using GrapheneOS for a long time already? Do you know if there are any useful-in-the-real-world local voice assistants? I mostly want it for setting timers and simple stuff.
Hey @simon I believe you would be interested in this? Pretty sure you were the one talking about how foobar doesn't have good audiobook-style speed controls... So @Bri managed to wrap speedy, a Google fork of sonic adding a nonlinear mode, into a foobar DSP! github.com/masonasons/FooSpeed…
Tak vám musím říct, že moje řidičská kariéra nejspíš právě dostala trhlinu, kterou už tahle firma nezalepí. Minulý měsíc jsem bral extra směny za odpadlíky, snažil jsem se pomoct kde to šlo a BAF, dnes payslip, kde mě firma podělala skoro o £500. Zítra to prý prověří a samozřejmě dají do pořádku. To znamená ve výplatě v lednu. Takže místo plánování dárků budu nasraný jak četník celé svátky. Stálo mě to 2 hodiny procházení záznamů z tachografu, a pekelného nasraní. Upřímně, teď bych s chutí rád prošel záznamy za rok zpět, protože tam zaručeně taky byla nějaká kulišárna, ale nemám na to mentálně kapacitu, protože bych asi vraždil
My new microphone and other audio essencials now are ordered from Thonmann! (thanks to @jonathan859 and Gemini for recommendations) Essencially, the entire thing started by me, who was spamming poor Jonathan with minutes of audio messages on Whats App about how I had no idea what to get myself for Christmas. Then I thought about my setup. My computer itself is up to date. As are my headphones and my keyboard. But then I realized that my good old Fifine k690 'already' is 3 and a half years old, and I thought that I could do much better with my audio interface. Back then I bought it primarly for Teamtalk and didn't make usage of the mono mode. However, as the years passed and use Discord more and more i realized how cool mono actually is, and unfortunately, the Fifine is not so great with that. ("Besides", I thought, "2025 I basically upgraded everything I own, so why not sort out the last piece of the old setup too?") So, I bought myself the Shure SM58 LC, the Focusrite Scarlett Solo 4th Gen Millenium MA-2050 and the sssnake SM6BK For those amongst you bwho aren't audio nerds (I'll include myself): * Shure SM58 LC: The actual microphone (the lc at the end means that this is the version without the 'switch' (what said switch is I didn't figure out either)) * Focusrite Scarlett Solo 4th Gen: This is the actual audio interface. For those who might be wondering why I need that, this is essencially the part which transforms the analog signal into a digital one a computer can perceive. * Millenium MA-2050: This is the microphone arm - so my microphone doesn't stand on the desk but hangs in front my face. * The sssnake SM6BK: This is just an XLR-cable to connect the microphone with the previous element. The total amount I paid is roughly €256. I am very excited when it arrives, and how it will be. I probably won't be making an unboxing or something, but I will gladly tell you about my experience.
Hopefully you'll have better luck with the Shure microphone. I find that most voices sound far too muffled, especially with dynamic microphones, so I don't understand why everyone is so crazy about Shure.
@radiorobbe I mean it's all right for the price. But I definitely get what you mean. I think my audio interface gives it quite a good sound thanks to the preamps, but that might not apply to every voice yeah.
I don't like wearing headphones in public places but I might have to start doing that while shopping. The Christmas music at the supermarket today was genuinely upsetting. High pitched warbly vocals, and one of those fake-out endings where the final chorus goes round and round a repeat again and again so you keep thinking it's ending but it's not OMG.
#chatmail does secure email federation that works. Using lean "relay" servers that enforce metadata-minimized end-to-end encryption.
There is a new little tool that checks whether two relays are interacting well for messaging, and how fast. You need a terminal window and be able to execute:
pip install cmping
you can then run for example:
cmping mailchat.pl nine.testrun.org
to perform a "ping" test for e2ee-messaging. Also see chatmail.at/relays
Chatmail provides FOSS infrastructure for interoperable, secure, speedy and reliable end-to-end encrypted messaging. Check out clients as Arcane Chat, Bots or Delta Chat today!
@leberschnitzel no, it depends on location, and maybe speed of dns resolution. Note that the ping is measuring the time from client (me) to one server, to another server and to the client, so travel over four network connections.
@radasbona Do you run some Unified Push distributor such as ntfy in background? If you can run Unified Push distributor, what does prevent running Delta Chat the same way?
If the problem boils down to the users allowing UnifiedPush client to run and denying the same permission to Delta Chat, then it is not interesting to spend development effort on it as the problem can be solved by allowing DC to run in the background. We do support FCM (besides Ubuntu Touch and Apple) notifications for the phones that do not allow to run apps in background, but there you cannot run UnifiedPush as well.
Yes. The pushes from DeltaChat have no identifying data/metadata, it just wakes up the app. Also the custom push server used for APNS/FCM scrubs the token from memory after the push is complete. Apple/Google may be able to see there was a notification destined for your device, but nothing else.
I think we could argue until the dogs come home about whether or not it's enough of an "established standard" when the number of users who refuse to use Google Play Services on their phones is dangerously close to zero and the number of apps available that implemented Unified Push is a rounding error when compared to the entire Android apps ecosystem
Delta has to target the common user first before it can expend the effort satisfying a very niche userbase
Thanks @Keonik1 for #614 :) I had a lot of fun with git rebase to break it up into small, reviewable chunks. It's still a lot, but doable :)
Before we can merge this, we need to merge:
cmdepl...
@d1 actually if you run fedora when you install "docker" from repo, it's not docker but the moby-engine (also made by docker but it's an open framework).
docker is used everywhere so they would touch more people by making first a docker image (compatible podman).
probably the previous posts were about running a relay, not just the "cmping" tool we highlighted in the top level post. all the custom code in relays is also done in python btw. Overview on relay architecture got reworked lately chatmail.at/doc/relay/overview…
@jarrodu @d1 yes but it do not remove my answer on the fact i have no intent to run something on the os itself, if no docker image (or equivalent containerized image) is available i will certainly never install it.
Because with docker image i can backup a docker compose and the file.
When I see a bunch of moving parts like that, I get cold feet, that it'll be as easy as people make out. Is there a Youtube/Peertube video on this? If it's as easy as you claim, then I'd like to see a from-scratch install on #Debian 12, using python's uv for the package manager of chatrelay (in a uv venv). Including sending initial testing messages proving it's working in production.
@d1 @jarrodu @lexinova Dusty, not quite sure what you mean with "tons of moving parts". It's true that we are aren't very strong in the "providing videos" game 😂 be our guest :)
@jarrodu main blocker for upgrading to Debian 13 is this issue: github.com/chatmail/relay/issu… if you subscribe there, you'll hear news when they come in. For now, we regard debian 12 as a fine target environment. If dovecot folks would finally merge our speed-patch github.com/dovecot/core/pull/2… then that would be a incentive for us to move to a new debian.
Dovecot 2.4 has compatibility-breaking changes, so configs will need upgrade: https://doc.dovecot.org/2.4.0/installation/upgrade/2.3-to-2.4.html https://web.archive.org/web/20250310165315/https://d...
@jarrodu @d1 and it's simply a nope for me, installing something on the system EVEN PYTHON, is a liability when you need an upgrade of OS, as a python update can break it.
Source code: Lib/venv/ The venv module supports creating lightweight “virtual environments”, each with their own independent set of Python packages installed in their site directories. A virtual en...
@teohhanhui @lexinova @jarrodu Thanks for that. I've used venvs before for a couple other python projects (copyparty being the most server-like). It becomes non-trivial when you want to set up a systemd service to start a server at boot time, and the venv must be invoked from systemd. Can systemd start, stop, disable, and enable the service reliably from the command line (systemctl)? Everyone apparently assumes these things are simple; hand-waving away such details as trivialities, but no they aren't trivial.
@d1 Chatmail relay uses systemd to run services in a venv. All the work setting this up is done by "cmdeploy" script, so you don't need to manage this yourself: chatmail.at/doc/relay/getting_…
As for "cmping", if you just want to run the command, you can install pipx.pypa.io/ and run `pipx run cmping example.org example.net`. pipx will manage creation of venv and setting up cmping package from PyPI.
I, for one, am waiting for them to support #Debian 13, not 12 (for chatmail relay). #Postfix changed their config variables as of 13; I'm concerned about an ugly upgrade path from 12 to 13.
Ok.. I love DC, here I have no problems with it, but why (without going into technical details, implementation).... On my phone, pure Lineage OS without Google services, without anything! Notifications are received on Telegram and WhatsApp, but there are always difficulties in Deltachat. Why! Just explain to me (maybe not even to me, in simple terms... for those who are very far from programming) why is there no simple implementation?! What's in the way?!
#France #administratif À savoir : Si vous devez fournir un justificatif d'identité, vous pouvez utiliser ce site officiel du gouvernement français. Il vous génère un justificatif d'identité à usage unique, en précisant le destinataire et la durée de validité. Cela évite de donner une copie de sa carte d'identité et en conséquence de moins risquer d'usurpations d'identité. france-identite.gouv.fr/justif…
Can anyone with Docker experience help us figure out what's causing our builds to fail? It was working a few months ago, and without any config changes it's now consistently failing.
A clean, Markdown-based publishing platform made for writers. Write together and build a community. - Merge pull request #1504 from writefreely/fix-customize-log · writefreely/writefreely@4b09f7c
1. Putin is sent to The Hague. 2. Russian soldiers leave Ukraine including Crimea. 3. Russia returns all kidnapped Ukrainian children. 4. Russia releases all Ukrainian prisoners. 5. Russia pays damages for everything their war has destroyed. 6. Russia pays damages to Ukrainian families who have lost family members. 7. Russia pays damages to every person who has been tortured, abducted or otherwise criminally mistreated. 8. Ukraine becomes a member of the EU.
[1]> Thank you for your patience and for continuing to share your concerns. We are listening, and we will get better.
I highly doubt it. Corpos rarely get better.
> Over time, accounts will enable more personalization, better reliability, and improved access to new features—including AI-assisted capabilities.
JAWS in educational settings on open computers do *not* need personalizing. Also where's that HIPAA and FERPA compliance? We need that for all this AI AI AI.
> But we should have explained this earlier. That is our oversight, and we apologize.
Lol corporate appologies mean nothing. Absolutely nothing.
> Some third-party AI integrations cannot legally be used by minors. Age verification ensures we comply with regional laws and provide the correct experience for all users.
And if we've turned off those AI features through the command line installer switches? No age verification needed.
> We should have explained more clearly that your information is not used for marketing unless you explicitly opt in.
For now. Later on they'll want that data.
> User Details for Adults (18+)
And here we go.
> Understand real-world workflows for people who are blind or low-vision
Dude surveys exist. They work quite well. Blind and low vision adults are, generally, capable of filling them out. They don't need to slow down an already unnecisary sign-up form.
> Allocate engineering effort responsibly
Oh they're already feeling the loss of top engineers.
> Phone Number
My number is, no. Besides I'm sure FS and their private equity bosses sell all this to data brokers. I mean that's what private equity wants, more money from their investments.
> We intentionally collect less information from young users to protect their privacy.
I wish companies cared as much about everyone else as they do kids.
Ooo lookie, infosec!
> Stored in AWS data centers with industry-standard protections
Industry standard isn't good enough. Shit gets hacked all the time.
> Encrypted in transit and at rest
Okay that sounds a tad bit better. But if it's at rest it can be decrypted.
> We have heard your feedback clearly, and we are making improvements.
Yeah every corporate says they've heard our feedback. So they should stop parroting that line and just lead with actions.
> Most fields currently required will become optional
Okay, over the next few weeks. We'll see.
> Only age confirmation and basic contact information will remain mandatory
Sure. Gotta get that phone number.
> Age verification will change to a simple “Over 18?” confirmation
Okay that's rather good I guess.
> Users under 18 will still be able to use the products but will have certain AI features disabled
So if you want no AI, just say you're under 18.
> Our Commitment to You Moving Forward
Bahahaha now this is straight up bullshit. FS is commited to making fucking money and nothing else. At all.
> The purpose of accounts is to make the products better—not to make your experience more difficult. We recognize we did not communicate this well enough, and we apologize for any frustration or confusion this caused.
See, they aren't backtracking on making every single individual user create an account. Oh no. They just wish they'd gotten control of the narrative around it before everyone else started looking at just how little info NVDA collects. Fucking assholes.
> Going forward, we are committed to:
Controlling the narrative earlier and more completely so our reputation doesn't take any more hits. Lol. Oh and going on podcasts to soothe the blinkyDink community.
> Working hand-in-hand with the community as we improve the software
They'll be on Double Tap soon enough.
Where I work, we have workstations where people use computers. There are lots and lots of people using each computer. Many of them, at first, don't even know how to type. Let that sink in. They cannot type. So how the actual crap are they going to make an account when they can't type? And no we ain't doing it for each and every student, and then showing them how to log out of their account and hoping they do that each day. Ain't gonna happen. Oh also HIPAA and FERPA compliance is pretty important where I work.
Meanwhile NVDA barely collects any information at all. Because NVDA is a screen reader. Also the AI Content Describer NVDA addon can use local, or even on-site AI models so no personal info leaves the workplace. Also NVDA is free. Also the on-site AI is free. Also the Gemma 27B model is fairly good at describing images. Also on-device or on-site AI doesn't require any account. Also none of this requires a subscription. Don't be stuck between JAWS. Get out before you get eaten. Unless you own a business yourself, in which case they'll probably treat you well, for now.
Learn why account information is requested, how it’s used, and what changes are coming to simplify Vispero’s registration process for JAWS, ZoomText, and Fusion.
Just for the archives, and to follow up on my recent post, it does seem to be possible to have two nvda remote sessions running on one instance of NVDA. I've got, before saying anything else, to thank @jonathan859 for the idea. It's amazing in its ease of use, and I am deeply grateful to him for suggesting it. I wanted to control two NVDA remote servers from one client. In other words, one laptop controlling two desktops, desktop 1 and desktop 2. I have NVDA 2025.3 running both the TeleNVDA add-on and the remote access that is built into NVDA. I have seen no issue in running both, the add-on updated fine and, of course, remote access just requires the checkbox to be checked. It is easy enough to get both to work. Just connect to desktop1 with the "remote" submenu in the NVDA tools menu or use alt+nvda+pageUp. Again, connect to desktop2 with alt+nvda+r, or through the "remote access" submenu of the tools menu. To control desktop1, hit f11. To control desktop2, go back to the controller and hit alt+nvda+tab. I am sure the keyboard control works well, but I still have to spend some time testing how the incoming text works when both controlled machines and the controller are trying to talk at the same time. It looks like it works well, too, at least I can't make it fail, but I'd like a few weeks to be sure. The annoying thing for me, and this is nothing to do with TeleNVDA, I think, is that I can't keep the computer connected through the remote access menu, the one built into NVDA, muted. I can mute it, but whenever I go to it with nvda+alt+tab, I get "unmuted". As I said, this has nothing to do with using two connections, it's just NVDA not allowing me to have keyboard control without speech output. Also, and this is even more of a specialized use case, it is possible to use one controller as a bridge to others. Connect to desktop1 via TeleNVDA, then have desktop1 connect to desktop2 via built-in remote access. As I said, this is usually not of any use, but can be needed when one computer is connected to two networks and can become a bridge between them. Again, thanks to @jonathan859 for the idea.
According to an email I just received from my utility company, "our app offers more."
More what?
Heartaches? Headaches? Exploitation? Bad news? Chances of winning the lottery? Unprecedented good fortune, the likes of which has never been seen before?
What are we comparing it to? Chain saws? Lawnmowers? Music production workstations?
Alright, it is time for a huge announcement, something that I've been wanting to get out there for a really long time, but decided to keep quiet about until it was fully prepped up and ready for showtime. Friends, I am happy to announce my first app release as an indie developer! Introducing Paylino, a simple yet powerful, fully accessible bill scanning and splitting app, available to download on iOS starting today.
Split bills in seconds — fast, easy, accurate.
Paylino is the iOS receipt scanner built for sharing. Snap a restaurant check, a shared grocery receipt, or any group expense. Paylino instantly detects line items, taxes, and totals, then lets you split everything fairly — down to the cent.
Start in a tap: Snap the receipt → auto-detect → assign items → choose tip → done.
How it works
1. Snap or choose a receipt Take a photo right in the app. Smart edge detection helps you line it up perfectly. Or import a photo from your library.
2. Capture everything automatically Our specialized AI recognizes the merchant, currency, items, and taxes/extra fees. Everything is clearly organized and fully editable.
3. Split fairly with friends Add names, tap who had what, and (optionally) select a tip percentage. Paylino calculates the total and each person’s share automatically.
4. Keep every receipt Your scanned receipts are saved in the app and synced with iCloud, so you can pull them up anytime.
For those of us who are blind / visually impaired, Paylino makes it super simple to scan receipts, and then check item prices and totals in a consistent, unified format that can be super quickly navigated with VoiceOver. I personally am really bad when it comes to lining up my iPhone camera for a good shot, so hopefully the fact that I've been getting excellent results consistently is a good indication that it should work great for others as well. If you've been reading this far and it's sparked your interest, I invite you to download Paylino from the App Store and give it a spin!
Note: since Paylino uses server-powered AI models for bill image analysis, using the app requires a paid subscription, the cheapest one is available at €0.99 / month, which enables 5 scan credits per month. Any thoughts, comments and feedback is super super deeply appreciated! #iOS #iOSDev #IndieDev
Entgegen den heutigen Schlagzeilen, denen zufolge die EU von der Chatkontrolle abgerückt sei, haben die EU-Botschafter heute in einer knappen Abstimmung den Weg für eine dauerhafte Massenüberwachung privater Nachrichten geebnet. Dr.
Contrary to headlines suggesting the EU has "backed away" from Chat Control, the negotiating mandate endorsed today by EU ambassadors in a close split vote paves the way for a permanent infrastructure of mass surveillance.
So, habe mal ein paar freundliche Hinweise an die Compliance Abteilungen von ein paar unserer Kunden geschickt. Schließlich sind uns unsere gemeinsamen Werte wichtig!
I really like this blog-post about AI, especially its last sentence: "The most disobedient thing we can do is to thrive." fokus.cool/2025/11/25/i-dont-c… #ai
I woke with my head buzzing. Not a headache; literally buzzing. I have chronic, loud tinnitus, and sometimes when I wake up, it makes these short, SUPER LOUD buzzes like a reciprocating saw cutting a tin roof. I don’t have a better way to describe it.
lain
in reply to feld • • •feld
in reply to lain • • •@lain I mean, I did explain exactly what I needed it to be able to do, which file to look in to add the functionality, and exactly what I wanted the configuration file syntax to look like and where the docs/man page was that needed to be updated.
and it did it. it just worked. I've read through the code about 5 times now and I don't know all the quirks of OpenSSL obviously, but there's nothing strange or scary going on
lain
in reply to feld • • •feld
in reply to lain • • •@lain i mean i guess if you open a new empty folder and say "Qwen, build me an MVP for an ecommerce website" and walk away you're going to end up with a giant stinking pile of shit
that seems to be how the average hater assumes the tool is used