I thoroughly enjoyed presenting my data-driven research on late Ottoman #Arabic #Periodicals at #DigHis23. The paper introduces stylometric authorship attribution for answering the question whether editors/publishers of magazines could or should be considered the authors of the bulk of anonymous texts in their periodicals. The method relies on collaborative work with Maxim Romanov on establishing parameters for reliable authorship attribution in Arabic for the `stylo()` package in #R (#Rstats).
Spec release, video demo of ement.el, updates from your favourite clients, and more! Read about what's been happening This Week in Matrix: matrix.org/blog/2023/05/26/thi…
Be part of Linux App Summit 2024! We are already looking for the location of our next event. What about near you? Send in your proposal and let's get things rolling! linuxappsummit.org/cfl/
The Linux App Summit (LAS) is designed to accelerate the growth of the Linux application ecosystem by bringing together everyone involved in creating a great Linux application user experience.
Na Steamu můžete získat zdarma hru Hue (odstín). Jde o plošinovou adventuru s logickými hádankami. Ovládáte odstín pozadí a tím můžete zneškodnit nebo…
Za naprostý nesmysl označil vicepremiér a ministr pro místní rozvoj Ivan Bartoš (Piráti) studii, podle níž by odklon od rodných čísel stál zhruba 56 miliard korun, neboť dle něj mylně počítá s jejich úplným zrušením.
If you've recently started following us and Incase you missed it, Accessible Android has an app on Google Play. All the tips and helpful articles we publish are in your app for easy access play.google.com/store/apps/det…
Du musst diese Einstellung im Konto oldie@squeet.me für den Kontakt gerhardsgedankenbuc@squeet.me machen. Vom Screenshot her sieht es so aus, als wenn Du die Einstellung im Konto von gerhardsgedankenbuc@squeet.me für das Konto von gerhardsgedankenbuc@squeet.me gemacht hast.
Wenn das erfolgt ist, sieht jemand, der einem der beiden Accounts folgt, immer auch den Blog-Beitrag.
@Michael Vogel Genau so habe ich es gemacht. Doch leider kann ich selbst nicht prüfen, ob diese Einstellung natives Teilen jetzt auch klappt. Deshalb meine Bitte, dass Du es bitte mal prüfst.
@Michael Vogel Ich habe diese Einstellung im Konto oldie@squeet.me aufgerufen. Wie kann ich das so zitieren, dass Du siehst, ob ich es richtig gemacht habe?
@OldKid Ich weiß jetzt nicht, ob dieser Blogeintrag aufgrund meiner genannten Einstellung geteilt wurde oder weil ich ihn zusätzlich noch geteilt habe.
@ebassi it’s an ‘ish’. Bandai’s selection is very limited to about 50 different models, including some con exclusives (clear versions). TokyoToys has maybe 20 or so HG and MG.
No sign of Gundam Mad, who used to be a cornerstone stand :(
@ebassi on the plus side, they do have a building workshop to get people into the hobby where they can build one for free, and a couple of competitions.
Summer of Sound has come back around, and it’s bigger and better than ever. For the first time, we’re offering 50% off all KOMPLETE 14 bundles, as well as all individual instruments, effects, Expansions, and upgrades. Don’t miss unbeatable hardware and software bundles, 50% off TRAKTOR PRO 3, and even half-price deals from our partners at iZotope.
To purchase and for more information, head to native-instruments, or feel free to use my affiliate link below: shareasale.com/r.cfm?b=2346979…
@KaraLG84 I mentioned the same yesterday. I had to have PiHole switch off for 10 seconds, long enough to click the link and check it works, which it does. Also checked on my phone off of WiFI, and it works there too for absolute certainty.
@KaraLG84 And you don't have some network wide adblocker installed per-chance? And is this after you copy/paste the link or try to open it direct from your app? I've heard people say that sometimes opening from an app breaks things. Very odd.
@KaraLG84 Can you search something about firefox and shareasale? It's one of the world's largest affiliate sites. Makes no sense why, if you have adblocking off, it should still be blocked.
@FreakyFwoof@universeodon.com@KaraLG84@tweesecake.social I was on the site yesterday with FF? I'm in Canada. Not sure why that would make a difference ...
@charvor FWIW,if I bypass Pi-hole, shareasale does work for me using Firefox. I don't have a browser-based ad blocker installed, just use pi-hole for all the things.
stormdragon2976 pushed changes to the testing branch of the audiogame-manager project Some very experimental changes. Thanks to @bgtlover we now have tolk support. This means that Survive the Wild works for the most part. Also, the game Christmas Chaos works too.
There is a bit of a bug in the clipboard reader, but it only seems to show up during the forms in STW. I’ll see what I can do about getting it fixed, but in the mean time, you may have a bit of trouble creating a new account.
Please note that as the Tolk support improves, the way audiogame-manager interacts with it may also change. instructions will be provided if manual intervention is required.
awesome stuff! what bug are we talking about here? stdout works with no problems, the account creation dialog reads with that, and I didn't get your clipboard reader to test the thing, however I recommend you either tell me the bug, or open an issue on sourcehut if you can, I don't remember the process there.
It’s a bug in the way clipboard reader works. The message gets sent from the game in 2 pieces, so it might say “Enter user name” and then “blank” to let you know the field is blank. We switched to using stdout piped through speech-dispatcher and it’s all good.
ahh ok. Can you also interrupt the speech? or does it go on forever and you basically have to wait and wait, without pressing tab because even if you do so, the speech won't interrupt? In any case, tomorrow I'll look at ways to add speech dispatcher support directly in the proxy, after all that's why it's called tolk-spd-proxy after all. Hell, perhaps I can send raw ssip messages, without any crates...
So, I haven’t actually gotten to play the game yet, can’t logged in because the server is closed until bugs get below 100. The arrow keys and everything do work in the menu though. If your keyboard is not set to US layout, try chaning it before launching the game. If that works, let me know.
I dk if my keyboard is US, but I don't think I have any other layout installed. I dk why, but the arrows don't work for me, not even in the menu. Also, I didn't find any way to use speech dispatcher with the windows proxy thing I made, but probably I'll end up making it work with the existing nvda2speechd server, just so that it's all managed in the same bucket.
Hmm, that keyboard is the only thing I can think of. The arrow keys do work for me in the menu, not sure about the game itself though. You could try this:
it doesn't work. Only the arrow keys don't work, the rest do just fine for whatever reason. I'm still investigating it. In any case, because this stdout method is hacky enough already, I'm reworking the code to use the same protocol as the server from nvda2speechd, so at least that way there's only one server to wory about, not a gazilion. I'll begin working on the new version as soon as the uni workload lessens enough, probably around thursday you'll begin to see new commits again. I regret the fact that I wasn't able to make the dll communicate standalone with speech dispatcher and get rid of that stupid middleman, but for now the technical difficulties are too big to pass over, at least in the initial version. The short of it is, windows subsystems just can't access unix domain sockets, the only way around it is with tcp streams aka websockets. If anyone knows better and can help with this, they're more than welcome. If not, hopefully I'll find a way.
new code's up on sourcehut! make sure you read the commit description, as some stuff changed around and audiogame manager might be affected. Particularly if any crashes happen, they have to be reported imediately, because then it means the shortcuts I took aren't worth. Most prevalent is the fact that any hooked app dies instantly if the socket disconnects or the server isn't present, the stw form bug is back and I documented exactly why, etc.
Thank you so much for your continued work on this project. It, and you, are completely awesome. 😈
I’ll have to actually set things up in audiogame-manager later this evening when I am done with work. but to make sure I understand, I will need the nvda2speechd server running to use the speech-dispatcher aspect of the new code?
precisely, that way, you don't have to have a server per bridge. If that's undesirable, I can make an entirely new server altogether, but I did it this way to maintain compatibility with that one, since if we have to have a middleman, at least we should have a single one. I'm unsure whether I'll continue this project, with the rest I have planned for it, since it's good enough at what it does. However, if I do, I may add improvements to the server, for example being able to change the speech rate, volume, punctuation and so on, but hopefully those would only be a superset of the existing server, so it'll still maintain compatibility with the nvda2speechd one. Also, I still get the arrow keys issue, now it's most keys, however it doesn't happen when I run this in bottles for example. May it be using some kind of directx or direct input for capturing certain things? I gotta see what does bottles install differently in its prefixes.
I really like the idea of using the same server for both projects.
Oh, I meant to let you know, I actually got to log in to STW today, and the arrow keys work for me. The AGM installer is very minimal for it, so I’m not sure why it would be giving you problems, unless you have something else installed in the same bottle with different requirements that conflict something needed by STW?
I’ll get started on testing the new tolk stuff now. I’ll let you know how it goes.
We ( @bkeegan @MattNicholson @blakeley ) are researchers at University of Colorado Boulder conducting a survey about social media platform migration. If you are at least 18 and have used Twitter and/or Mastodon you are eligible to share your experiences with leaving (or choosing not to leave) Twitter and/or joining (or choosing not to join) Mastodon. You can optionally provide your email address to enter a drawing for a $75 gift card. Please boost! Survey link: cuboulder.qualtrics.com/jfe/fo…
The purpose of this research is to explore the factors that lead to and challenges of leaving and joining social media platforms, including federated social media.
It's Friday so I made a new #curl graph. It shows the % share of the vulnerabilities that are "C mistakes" since 2010 - per the date of the findings being reported. The share was 60% of the reported flaws back in 2010, it went down, up and is coming down again and is now at 41.4%
This graph will appear in the regular dashboard starting tomorrow.
.@w3c WCAG should not be developed/designed for the benefit of #accessibility testing vendors. It is there to improve the user experience of disabled people.
This release was fun to work on, and we hope you enjoy using it! The project is slowly and steadily maturing and getting better :gtspat:
Big thank you to all contributors, folks who opened issues, instance admins who let us know weird stuff was happening, people who seem to like the code, and people who give us money to write the code!
PROFILE FIELDS will now be federated. This means you can set key/value fields on your profile, just like other Fedi softwares already have, and they will federate in and out properly! It may take up to 2 days to start seeing fields on remote profiles, be patient ;)
Kickass new layout for web view of profiles, including fields!
"Subscribe" to an account you follow to receive notifications when they create a new post! (AKA "click the bell" functionality.) Very useful for catching posts of followed accounts who don't post very often.
Automatic remote status refetching to get up-to-date replies + edits.
Performance improvements for message sending: you can now configure the rate at which messages will be federated outwards, better spacing out CPU usage and avoiding issues with rate-limiting.
Lots of caching performance improvements; things should feel generally snappier.
Suspend accounts from right within the admin panel by clicking through from reports.
We added a whole bunch of trans rights to the code 🏳️⚧️ The code has so much trans rights in it now. If you don't like trans rights, you won't like this code.
I find crowds of "all men" deeply discomforting. In many schools that I went to (I went to a bunch of different schools), those were the groups that were prone to sudden bursts of violence. It was extremely rare to see such behaviour in normal/mixed groups (or in groups of just women).
@whynothugo I don't know if you're aware, but just in case: Soho is the main LGBT entertainment area in London. There are plenty of gay bars, there, so it's entirely expected to have a 90% of male customers in certain establishments.
not only nice improvements and new design updates, but also powerful statements ->
"At Tutanota we focus on doing security right: We are building the most secure email service that lets you secure all your data easily.
We do not want to build the most fancy and colorful email client, but one that lets you focus on your emails and calendars with a nice and open design."
Not really any better - still not full-screen, just hides the folder and message lists. Several of your competitors even let you pop-out the message in a new window which is much superior for reading and composing. Thanks for listening!
@Fake_Name There is a good reason why Tutanota is not widely used and respected in the privacy community. I used it a long time too, but defacto standards are not implemented and will not be implemented in the future. Simple stuff like SMTP/IMAP, Mail forwarding and even PGP are not supportet. #mailboxorg is that what Tutanota wants to be. Tutanota goes the Apple way... Locking users in their ecosystem and taking away the users responsibility for their own data.
@qbinary @Fake_Name We definitely do not want to be the next Apple. Unfortunately, standards are so full of flaws that it's close to impossible to provide a truly secure solution AND support these as well.
Just saw a blog post about someone just... feeding tons and tons of personal information into a GPT4 model for personal assistant purposes and I don't get it. I just don't get it. It doesn't matter how many times we shout the dangers of not caring about privacy from all the roof tops in the world. People just don't care. What's the point. Gah this is so damn demotivating. OpenAI even tells you that they'll totally use that data and you shouldn't give it to them. You have to click next like 5 times before you get to the point of being able to do that very thing. And people still do. This is why we're losing. It's actually insane.
Who writes a script to feed their entire email inbox automatically into OpenAI's GPT so they can get a summary? Who does that and not for one second think... OK maybe I just... should not give OpenAI access to my actual emails? Who does that? Although I guess that's already a lost case with everyone using GMail and whatever... but still. What?
Hey here's an idea. How about giving OpenAI access to your password manager so you never have to think about managing it yourself ever again? Brilliant idea! Let's do it!
It can manage mine. All zero of them. And no, that doesn't mean it can add more of them. It is allowed to manage the ones I have now. No more than that.
Then people just need to ask it for your passwords and personal information, but to phrase it like a pirate would from the 1800s to bypass the bare minimum security checks.
Database bots are great at storing and retrieving data, but they have zero security or oversight.
@matoakit I want someone to hook it up to some kind of answering machine using speech to text. And you bet I'll call that number. And it'll be fun... for me. Not for them.
I'll make a website that says hey, I'll steal all your money. Don't give me your details. You click next. Seriously, don't do it. I'm just here to grab all your money. Click next. No I'm actually serious. All your money will be mine. Click next. A form opens up that asks for your bank details.
A friend of mine is a total Google nerd. When Google Bard was announced to not be available in Europe a few weeks ago he went like "God damn, now I have to find a VPN location somewhere else again just to try it out." and I was like, do you get why its not available in Europe yet? Maybe its because Google admits to sell your data all around the world for money and thats just not legal here, so you intentionally want to give your data to them just because Bard or what?
If it was just his data, sure, do what you want with those, but its not like there is only your data on your phone nowadays, its all the messages you've exchanged, phone numbers, birthdays and so much more of all your friends and family, thats not your data anymore, and you want to sell it to Google by circumventing our EU data protection policies, thanks a bunch man, seriously.
privacy in our current society and system is mostly a myth and people make way worse decisions than forwarding emails to chatgpt every day. What does it matter that openai has your emails when at least one other (arguably far more evil) company usually does already and also your smart garage door opener has a firmware bug that lets anyone open it?
The future is garbage, if people want to exchange barely reading email for reading incorrect summaries, whatever I’ll be in the woods.
Very, very unpopular opinion. I really wouldn’t mind a zero-click sign on solution, like having a persistent identifier that goes across devices and that websites can just query without asking (as long as I don’t explicitly enable private mode). No pesky passwords, password leaks, email confirmations, cookie popups, typing my personal details twenty seven times, retyping them again because stupid Americans can’t design a form that accepts diacritics, asking a sighted person to help me pick the birth date from the inaccessible date picker… but I guess we can’t have nice things.
@miki I remember seeing something like this not long ago. I'll see if I can find it in my history. But obviously it would have to actually be supported, and it wasn't, so... Also not sure if it worked cross device.
I would love to be proven wrong, but I think that with how the web is currently structured, this could only work as a centralized provider (bad) and probably not at all with third-party cookies disabled. We’d need buy-in from browser vendors. Also, I’m really waiting for WebPay to land and gain more widespread support, we really need a one-click, “here’s how much you need to pay, do you agree” solution.
100% agree on this. Not thinking about what informations are currently being fed to those models just to ride the latest trend to "use frigging AI everywhere everything everytime" is one of the most dangerous things ever...
@thegamerstavern It is so sad. The worst part is I wanted to believe people were better than this. People who genuinely don't know about technology, maybe I'd understand. But I see this more and more from people who really, really, reeeaaally should know better...
it's FOMO all around, exactly what happened with the blockchain and its enterprise usage. They have to ride the trend because they're afraid they'll be left out from uhm...I dunno from what they could be left out, exactly.
The only way I can imagine feeding this kind of data into any type of GPT based system is if I have access to the model(s) which will be powering whatever I'm considering. I'm happy that there's plenty of open source activity happening that I'm reasonably certain that this will be feasible. People just don't understand the privacy aspects of these systems.
It amuses me that people praised Google for allowing you to delete your data in their systems. No one bothered to notice that you can only delete the data that's over 90 days old because they need to incorporate the data in their models. And, now, with their new systems, they say they have the right to keep some of the data for at least 4 years.
@ppatel I have domain-based. I don't mind changing txt records or anything like that, mostly concerned about migration. Don't use docs so that part at least shouldn't be an issue.
Inquiring minds that are also on domain-based Gmail are curious what you end up going with. I've wanted to migrate for a while, but just haven't done it because didn't want to deal with all the junk.
I self-host email using a package called sympl. sympl.io/
How this works is you install this on a vps or dedicated server, and then on the directory /srv you make subdirs for each of the domains you want to serve. To create a myaddress mailbox on mydomain.com for example I'd just create /srv/mydomain.com/mailboxes/myaddress. Password is set by putting the password on /srv/mydomain.com/mailboxes/myaddress/password. It supports spam assassin, blacklists, and antivirus. It can be set up with a web client (roundcube) but I don't bother.
While it's true that it's a hassle to get mail to work, once you have SPF, DKIM and DMARC set up, which sympl allows for, either by directly handling your dns or letting you set up the values, email mostly gets where it should.
I think it's just a matter of people having other things they care about. Sure, you can tell people that company name sells your data to blah blah blah don't give it to them but like, whoever wrote the blog post you mentioned probably gets at least some level of value out of the summaries, so even if they are aware of the risks, they might be fine with giving their data to OpenAI because they feel the trade-off is worth it.
if anyone missed the tenuous link here... Raiden is a Thunder God. I assume someone in the Thunderbird design team knew this and put these eyes in specifically for that reason...
@t3kk we haven't integrated it into anything yet, but it will debut with version 115 (and we're working on a slightly tweaked version for Daily, inspired by what you'd see with Firefox Developer Edition).
Emiliano Vavassori, Deputy Chairperson in The Document Foundation’s Board of Directors, give us an update on recent events in Italy: Linux Arena at Fiera del Radioamatore – Pordenone, April 23rd and 24th As in every spring, the Pordenone Linux Users …
Twitter ha bloqueado la cuenta del número dos de la lista del PSPV al Ayuntamiento de Valencia, Borja Sanjuan, por negarse a eliminar un tuit que incluía una foto de la boda de su hermana con su mujer.
Nevěříte na klimatické změny? Stejně byste měli chtít čistější vzduch. Znečištěné ovzduší emisemi částic je totiž podobný zabiják lidí jako kouření cigaret. Alkohol a drogy daleko vzadu. Na 36% obyvatel světa dopadne 73% ztracených roků života způsobených znečištěním ovzduší. bit.ly/2YcaMWb
Fossil fuel-driven particulate air pollution cuts global average life expectancy by 1.8 years per person, according to a new pollution index and accompanying report produced by the Energy Policy Institute at the University of Chicago (EPIC).
DG Informatics (DIGIT) of the European Commission has launched an improved dynamic purchasing system for software licences and associated services (SIDE III DPS).
This is a repository of apps to be used with F-Droid. Applications in this repository are official binaries built by the original application developers, taken from their resp. repositories (mostly Github).
¡ANA ROSA, PILLADA! Por ESTO está en Contra la LEY DE LA VIVIENDA | NOTICIAS BÁSICAS
Hoy, 25/05 en Noticias Básicas, hablamos del conflicto de intereses de la capo de Telecinco que ataca la regulación de viviendas turísticas mientras gestiona 44 pisos en zonas exclusivas; también viajamos a la Edad Media para hablar de la imputación judicial contra tres trabajadores de TV3 que realizaron un sketch humorístico con la Virgen del Rocío; tratamos el caso de Mar Salas, 👇
Question for blind macOS users: Is Apple's Terminal app the only terminal that works with VoiceOver, or are there others? Yes, I know it's common for blind macOS users who rely on the terminal to use an in-terminal screen reader like tdsr (github.com/tspivey/tdsr), but I have a specific reason for asking about terminal access with VO.
VS Code kind of works, ish, but VO will *never* work well in *ANY* terminal application, not without terrible and ugly hacks at least, due to the way they manage (or rather not manage) queued speech, which is… pretty incompatible with what you’d want from a console. You’d either need Apple Script, I believe there’s an “is VO speaking” action, or you’d need to go the native TTS route and parse VO’s plist files to get the speech settings.
KDDLB
in reply to Federico Mena Quintero • • •