If you're helping people sign up on here, that's great! 👏
However, I'd strongly suggest not signing up on the mastodon.social server. It's way way too large and that's dangerous (here's why fedi.tips/its-a-really-bad-ide…).
I know some find choice overwhelming, so if you just want one general long-established very busy server mas.to is a better option for example.
FWIW I'm reasonably sure Facebook never supported XMPP federation. They supported logging in via XMPP with a regular XMPP client (c2s) but they didn't support s2s. Meaning you could only chat with other Facebook users.
GTalk had federation and it was somewhat bad when they shut that down. But I don't remember Facebook's XMPP access being very relevant for the larger ecosystem.
@daniel I agree that it would be a better example. Wikipedia has decent references if you'd like to update the article: support.google.com/talk/?hl=en…
Také zdravím, a omlouvám se, že mi to tentokrát nevyšlo, ačkoliv jsem to napoprvé zaškrtnul. Člověk se občas vzdálí moc daleko od domova a takto to dopadá. 😄
I just had to uninstall vipre advanced security because sometimes when I would reboot the computer, my whole internet on my desktop wasn't connected. I would reboot, and it would fix its self, and then if I had to reboot again, it would break my internet again. I don't have time to try to figure out why in the sam hell it would break my internet that bad. I did try turning the firewall off, checking for internet, and there was nothing, and turning it back on, and still nothing. sometimes when I would go check, it was already turned off, but I had no internet at all on my desktop. That was fun while it lasted, but I am not paying for something that decided to sometimes cause me to have absolutely no internet access when I reboot the computer. back to microsoft defender I go I guess.
ah sorry to hear! It's working great here but I've also added like 40 or 50 firewall rules by now, my .XML file for firewall rules is almost 380 kilobytes large! Holy cow! So it's a lot of constant rule adding. Their firewall alone can be turned off in the panel too if you don't want it and just the virus protection, but it's definitely an aggressive one! Else you can always just tame defender if background scans bother you from it, it's somewhere in scheduled tasks from what others and the web have mentioned, there's one for the scan schedule itself.
@Tamasg I don't know how many firewall rules I had created, but it felt like over 50. I know office had several alone, and same for one drive. if it wasn't for the fact it was doing that to me, meaning killing my internet connection like it was, I would have kept it. oh, btw, not sure if you found it or not, but I can deffinitly tell you you can change how long things stay in quarenteen. you can do 2 days, 1 week, 2 weeks, 1 month, 2 months, or never delete. I did see that this morning when I went to look around at things.
ah yeah, but you can't tell it to not auto-quarantine sadly.... That's more what I (and I think you maye as well) was looking for. It just throws it right away into it. Defender does ask you if you want to allow on device, but of course you have to catch that notification in time to do it too. So, ups and downss I guess. xD
@Tamasg and sadly when you try to look at what was in quarenteen, you couldn't see the file that got nailed. at least I couldn't, and I tried NVDA, and JFW. in fact, NVDA read more of that than JFW did. I think the same is true for the firewall rules. with the firewall rules, if you listned long enough to a rule, you could hear something like C:\portable\Tweesecake\tweesecake.exe, but I didn't hear anything like that with JFW. Also I did run into an interesting thing with JFW, and Leasie when the firewall hit them, and was asking me to allow them, and what not. things were seriously haulted for a few minutes, until Leasie failed at activating or varifying on JFW startup. once you got past that point, you could create the rules, and keep kitting allow on all of the JFW stuff, and then from then on it worked as it should, but just the anitial of it was painful. when I say painful, I mean you couldn't do anything on the computer it froze things that bad. smile.
@lino_jr @vol4life8657 @Tamasg the thing is I want *completely* control of *everything. I want the popups that ask me if I want to create a rule for something and then allow it or deny it. I don't want software *ASS-U-ME-ing for me.
@lino_jr @Tamasg Well avast will give you all the pop ups you could ask for. ANd I"m not saying that to be mean, but it will also let you change things in settings, to give you the ability for it to ask you every time before it takes an action.
@vol4life8657 @lino_jr yeah avast is not a bad option actually. I think that the smart scan thing can also be disabled in it, and you do have a lot of control. Who knows, might switch, not like I don't have 30 days to cancel this one. Well now like 28. LOL. Still, I got tired of Defender's inability to customize scan schedule and what happens for so-called "catch-up scans." Like when a scan is interrupted in the background by a reboot and then what. I don't want it continuing then, I want it to restart the full scan. things like that I need perfect LOL
@Tamasg @vol4life8657 @lino_jr I know personally I don't like things going into quarenteen, without me telling you, hey! ignore it, or something like that, and I want to be able to read my quarenteen, which unfortunitly you can't do with viper. you only hear the what it thinks is the virus name, but not the filename.
The EU Regulation to Prevent and Combat Child Sexual Abuse, first published in May 2022, has become the "most criticized law of all time". Here's how to stop it now!
This is not a European-only problem. It is a global issue. It would take almost no time for them to move from that to banning software not connected to a cloud service, which is further connected to a digital ID of sorts.
This means Linux would become illegal in its current state, plus LibreOffice, OnlyOffice, Blender, etc.
Chat control could "end the right to privacy in Europe"? Sounds like hyperbole.
If it's a bad idea, say that. Or, that it will reduce privacy ...or that it could expose at risk people. But c'mon "end the right to privacy" is a long way down the road from what chat control does.
fair.tube/w/72DCPMByyS1hoSg7gQ… Here a good video explanation of the Max Plank Institute on how the technology of Chat Control works and what this would mean for digital communication within Europe.
I've tried all the new-fangled web design apps, and I’ve decided to stick with building with raw JS, HTML, and (S)CSS. There's just no need in most cases for all the cruft those apps generate. These days you don't even need big JS/CSS libraries for most things. I can make the same page 1/10th (or less) of the size of what visual web dsign apps are going to make.
I don't want to sound curmudgeonly about it, but I think these apps are making the web a little worse. It's starting to feel like we're back in the Dreamweaver days.
So sorry to say this, but NVDA's new on-device image captioning (introduced in today’s alphas) is consistently unreliable and basically useless. I get the security concerns behind keeping it on-device, but the model has to be useful to be viable. Here’s an example. I tested an image, and NVDA gave me this short, irrelevant description: “A blue and white sign that is on a blue wall.” Meanwhile, Be My Eyes AI gave me this: “Two hikers standing on a rocky, mountainous trail with a dry and rugged landscape in the background. The person on the left is wearing a red and black jacket, green pants, sunglasses, and a grey backpack. The person on the right is wearing a blue long-sleeve shirt, dark pants, sunglasses, a patterned bandana, and is holding a hiking pole, also carrying a backpack. Both are standing next to a pink metal sign written in Persian (Farsi) script. The sky is partly cloudy, and the surrounding mountains stretch into the distance. It appears to be a sunny day.” @NVAccess
Oh gosh this is already in Alphas? That feels more like the models used in the early days of local-image recognition, perhaps they're keeping model size down to save space. One option could be to offer cloud-based downloads for larger image models but of course then there's bandwidth questions, nonetheless it's for a balance to ponder. @amir @NVAccess
@serrebi I mean, you can't run something super powerful without including a bunch of runtimes like ONNX and all that, so perhaps in a way I feel for them, as well as NVDA used so widely across processors that you need to support from N100 low-powered Atom-class to high-class I9 and Rizen 9. For the latter a larger quantized model would not be a bad choice and would do way better in output but then you contend with it not working well if at all on low-CPUs. So glad this is still alpha though we have time to iron these things out. @amir @NVAccess
I agree. They should support a remote AI , for those of us not concerned as hard about privacy. Obviously don't scan really private stuff, that's a given. I think the main point of the feature is scanning UI, is it that bad at that? I dunno, if you were gonna implament AI at this level, you'd think it would be used for object labeling, etc. I'd look, but I don't want to live without IBMTTS. I'm also kind of concerned that this feature is too early: Like windows might have an official pipeline for using different AI models soon. Why not just use that when it comes out? Maybe that's prognostication, but I'd expect something like that from windows soon.
@Tamasg I don't even think this is an LLM at all, these descriptions sound more like what VoiceOver recognition uses, simple object detection from like four years ago, way before LLMs could even describe images. Even a small VLM, like Moondream or Gemma3 would produce far better results than this.
@ZBennoui Exactly my thought. It really feels like old-school object detection rather than anything close to a modern VLM. Even lightweight models today do far better. And honestly, if a feature is added, it has to provide at least some real usefulness. Otherwise, what’s the point of shipping it? @Tamasg @NVAccess
@Edent Hey! More then two years ago, I contacted you to setup an import tool to import all the benches from OpenBenches into OpenStreetMap, but I got stopped by a missing feature in an upstream dependency.
Great! I'll update the challenge in the coming days.
The import is setup to add the OpenBench-id into the OSM-object with `openbenches:id`, as described on our wiki: wiki.openstreetmap.org/wiki/Ke…
I've asked permission to the community (here community.openstreetmap.org/t/…) to already do an automated edit where all benches in OSM that already have a link to OB, will also receive the data (and images) that are in OB.
Hi everyone, https://openbenches.org/ has clearly indicated that they want to integrate their data into OSM. Their license is CC-BY, I [asked them] to clarify how they see the attribution requirement.
@hub Really? Ugh. Kevin Spacey lost work and he's still working in that field. I used to love these movies, but they have been rehashing them so much, that I've become bored with them. And this does not encourage me to take another look.
Download the most advanced FREE DJ software available, featuring iTunes integration, MIDI controller support, internet broadcasting, and integrated music library.
There's nothing surprising about Apple removing ICEBlock from the App Store. Much as we might want companies to take moral stands, they're simply not designed to—that's the system we've built for ourselves. sixcolors.com/post/2025/10/app…
So, I guess I’m going to need to keep writing this piece. Say what you will about Apple’s decision this week to remove from the App Store the ICEBlock app that provided a way for people…
I was able to use an extended conversation with an AI chatgpt.com/share/68ded9b1-37d… to help answer a MathOverflow question mathoverflow.net/questions/501… . I had already conducted a theoretical analysis suggesting that the answer to this question was negative, but needed some numerical parameters verifying certain inequalities in order to conclusively build a counterexample. Initially I sought to ask AI to supply Python code to search for a counterexample that I could run and adjust myself, but found that the run time was infeasible and the initial choice of parameters would have made the search doomed to failure anyway. I then switched strategies and instead engaged in a step by step conversation with the AI where it would perform heuristic calculations to locate feasible choices of parameters. Eventually, the AI was able to produce parameters which I could then verify separately (admittedly using Python code supplied by the same AI, but this was a simple 29-line program that I could visually inspect to do what was asked, and also provided numerical values in line with previous heuristic predictions).
Here, the AI tool use was a significant time saver - doing the same task unassisted would likely have required multiple hours of manual code and debugging (the AI was able to use the provided context to spot several mathematical mistakes in my requests, and fix them before generating code). Indeed I would have been very unlikely to even attempt this numerical search without AI assistance (and would have sought a theoretical asymptotic analysis instead).
I've been comparing the sequence of the Least Common Multiple of the first $n$ integers, $L_n = \text{lcm}(1, 2, \dots, n)$, with the sequence of Highly Abundant Numbers (HA).
The two sequences in
VIP Customer: Build our new office network/tech/CCTV/PACS/BMS infrastructure, possibly with some server capacity... never know if we need a couple AI rigs.
Me: Okay. It's weird how the building has 5 electrical meters but one tenant, and maybe we should add some EVSEs too.
Cx: Don't worry about the meters thing, and yes, EVSEs are a good idea.
Me to Elechicken: How much power (KVA) can we get out the existing panels? What voltage? What current? (So I can order UPSes & PDUs.)
Elechicken: (getting back to me) Uh, I have to call the electrical safety authority because the building isn't wired properly [no disconnects on 600->208 secondary, times 5]
Brits complaining that they can't engage in protests "against Israel" for one whole weekend after a synagogue was attacked and Jews were killed on the holiest day of the Jewish calendar.
These protests usually contain calls for violence, antisemitic tropes, antisemitic rhetoric, and many times "just happen" to be outside of Jewish schools, synagogues, inside secluded Jewish neighbourhoods or even Jewish restaurants- never outside the Israeli embassy.
@hub ahaha, thanks for that! I was going to go digging but didn't have time at the moment
the last things I found were that they were supposed to build a test track in Droux that hasn't happened yet, and they created a 1/3 scale demonstrator that didn't work 😅
@somecanuckchick in that case we observe the level of incompetence. Instead of building something that is known to work, they want to build something that is not known to work.
@hub Whenever I hear/read someone working for exposure... I am reminded that exposure is a CAUSE OF DEATH, at least where I grew up in Ontari-ari--ari-o. @shadowspar
Nicholas de Santo is a great comedian. He dare to say things lots of people are scared to say about modern politics and especially religion. youtube.com/watch?v=uxQZdZwOOH…
Time to honour the "Islamophobia Awareness Month". As the UK launches this smart initiative, here is a compilation of Islam-related jokes. Are you aware yet?...
Indeed it is easy to see that an odd-dimensional sphere admits a non-vanishing tangent vector field through a simple process of considering coordinates of the ambient even-dimensional Euclidean space.
Today we merged a new feature in curl: support for Apple SecTrust. This will become available in curl 8.17.0, due to be released on the 5th of November 2025.
tl;dr - I made a tool to look up WCAG SCs and copy the links. As anyone who has spent more than 30 minutes with me where the conversation might veer towards anything technical, I am an über-bore where it comes to the Mac app Keyboard Maestro (KM).
We've spent days in hospital and the poor wife's had a lot of things going on, but not the things she wants to have going on. Baby is still firmly ensconced.
The house is coming together well though. The first of the sofas appeared yesterday, we put the curtains up this morning and the carpets are all down and lovely underfoot. The window sills have all been painted and varnished, the French doors are sealed and watertight (a good thing with a storm blowing in), and I've just finished re-seating all the electrical wall outlets and light switches now the paint's all dry.
The baby's changing unit remains to be assembled, but the crib and car seat are ready. We just need a baby now. Daily CTG's are strong. Baby's moving well internally. Things just, have yet to go pop!
is it normal in the UK for her to be in hospital this early? Here they won't admit in most cases until contractions are more than like 8 an hour lasting a minute or more, or water has broken, or dialated above like 5cm.
@dhamlinmusic not typical, but they wanted to induce labour early due to diabetes. They tried, but before we got to the water-breaking stage there was a mad influx of emergency sections (they've had over 500 deliveries in September). On our second morning there they had 14 cases booked to come in for induction, and 10 of their 10 beds were already full from previous inductees who hadn't moved on. We've just had the call up now to go in for the Amniotomy in a few hours.
feld
in reply to Dev • • •Dev is back? In Redmond?
(Hey it rhymes)