If you’re blind or visually impaired and have struggled with taking notes or recording meetings, the Plaud NotePin might just be the game-changer you’ve been waiting for.
Soon, in the first weekend of May 2025, we are gathering curl enthusiasts in a room in the wonderful city Prague and we talk curl related topics over two full days. We call it curl up 2025. curl up is our annual curl event and physical meetup.
- Matrix Gateway support for rooms - Security improvement with ability to store multiple simultaneous password types - Ability to execute API commands from an XMPP client
and more !
This is a major release and you can check the details for the release on ProcessOne blog:
We are pleased to announce ejabberd 25.03, a big release with new features and many improvements, as it comes three months and 180 commits after ejabberd 24.12.
Completely rewriting a multi-million line COBOL codebase that has life-or-death consequences for real people in the space of a few months, using gen AI?
I’ve been writing software for 40-some years, and I have to say: this may be, without exaggeration, the stupidest software-related idea I’ve ever heard from leadership.
Here's an Acorn Atom, released in 1980. I haven't ever used this computer myself. But neither have I used an Apple III, which is, as far as I know, the other well-known computer that came out that year... well-known for being a dud. The photo is from Wikipedia.
As I'm visually impaired, I'd appreciate a more detailed description of this image. #ALTforMe
@matt Description: well, have you ever seen/touched a BBC Model B computer? The Atom is its predecessor. Case is the same flat box with slope around the keyboard, but white plastic with grooves running front-to-back behind the keyboard. Keyboard is the same layout but no red keycaps or function keys. Logo says ACORN ATOM instead of BBC Model B. Looks to be the same size.
Hi there, working on moving old friendica sites from a shared hosting to a VPS, I ran into the following problem I can't find a solution for. This actually happens with one 2021.04 instance but not with another (anymore), so I'm crosschecking all the settings but can't find how to solve this. Both sites were installed as 2012.04 from the start:
not working version wrote:
This is Friendica, version 2021.04 that is running at the web location sub.domain.com/var/www/subdomaincom. The database version is 1413/1413, the post update version is 1400/1400.
working version wrote:
This is Friendica, version 2021.04 that is running at the web location sub.domain.com. The database version is 1413/1413, the post update version is 1400/1400.
As visible in the screen, the site also displays the following path on top of the page, and of course calls that URL: sub.domain.com/var/www/subdomaincom
Different intents to fix this by changing or deleting the following settings from the console didn't work out:
Even more astonishing, sending the following setting with the console: /bin/console config system url https://sub.domain.com .. the site is displayed and works as it should be for a few minutes, and than falls back into the wrong behavior, like if the worker changes something when ever the cron kicks in. In that time span I can log in and even go thru the admin settings but this setting is not available.
Intents to modify the following specifications in local.config.php didn't work out either:
At least I figured out/ realized that url. and system. actually shouldn't be used at all as:
'config' => [
],
'system' => [
],
.. are actually "in charge" of defining that. So now I'm not sure either if those wrong lines somehow made it from local.config.php into the DB and are now part of the problem.
and it actually looks like this is exactly te problem but I have no experience at all with mysql over CLI SSH and don't know what commands to use to delete the apparently wrong setting in the DB.
Is there a way to delete entries with bin/console? I did manage to change some settings from there but couldn't find any hint how to "delete" an entry entirely from there:
@utopiArte The last command looks correct, although this kind of low-level config value should be set in the config file, not with the console that only interacts with the database config.
Trying to document changes and results in kind a step by step answers. The following setting leaves the site with a blank page and results in bin/console config printing the error message: [Error] Missing database structure config file static/dbstructure.config.php
Site doesn't work, resulting in the screen as shown in the inicial post. Setting the URL as described: /var/www/inversioncom/bin/console config system url https://inversion.tupambae.com .. the site works for a few minutes.
Site doesn't work, resulting in the screen as shown in the inicial post. Setting the URL as described: /var/www/inversioncom/bin/console config system url https://inversion.tupambae.com .. the site works for a few minutes.
Site doesn't work, resulting in the screen as shown in the inicial post. Setting the URL as described: /var/www/inversioncom/bin/console config system url https://inversion.tupambae.com .. the site works for a few minutes.
On a first glimps, I haven't found any mention of the path terms. In any case I will wonder about the meaning of: "the meaning of the relevant config keys". Like to say, if I can find "relevant config keys" ..
@utopiArte Sorry, I was on mobile and couldn’t check but these keys should have been mentioned in this file, along their purpose.
system.basepath is the absolute file path where Friendica is installed, e.g. /var/www/subdomaincom in your case.
system.urlpath should only be used if Friendica is installed in a subfolder of the domain you’re using, and is relative to that domain. It doesn't seem to be relevant to you, so it should be blank.
system.url is the absolute URL where your Friendica node is accessible. It should consist in the protocol, the domain and the urlpath together. In your case it should be https://sub.domain.com/
system.php_path should be the absolue path to the PHP CLI executable. It should not point to PHP-FPM, which is used to process web requests.
Changed this setting in the local.config.php to phpX.y. The sites are working like before.
I decided to copy/paste identically local.config.php of the site that does work to the one that doesn't, only changing the specific detail of DB and folder structure, and also made sure that the .conf files are alike. Still having the same problem.
Compared the print of: /var/www/subdomaincom/bin/console config in the few minutes it is working correctly after sending: /var/www/subdomaincom/bin/console config system url https://sub.domain.com and when it doesn't work but the prints are identically.
The print right now for both sites show the following quite at the bottom of the print:
The only difference between the two sites in the print is the following line quite at the top: config.system.basepath => / This line doesn't exist in the site that is working and I can't find a way to skip it or delete the /. Meld (the diff program) shows a dot in front of the the slash ./, the simple text editor Kate doesn't.
Meld actually also shows the [enter] signs in the same way, a little bit less colored, in lines that have them, like it shows the dot. 🤷♂️
(notes (to myself))
hypolite wrote:
system.urlpath should only be used if Friendica is installed in a subfolder of the domain you’re using, and is relative to that domain. It doesn't seem to be relevant to you, so it should be blank.
As of now local.config.php does define system.urlpath, but as the site that works uses the line I left it for now.
hypolite wrote:
system.url is the absolute URL where your Friendica node is accessible. It should consist in the protocol, the domain and the urlpath together. In your case it should be https://sub.domain.com/
system.url is actually not set as https://sub.domain.com/ but as sub.domain.com
What do you mean by "works for a few minutes"? How many pages correctly show in the mean time? What happens when the few minutes have elapsed?
The site works correctly for several minutes. I can log in. All pages show correctly, including admin pages. Once it falls back to redirecting to: sub.domain.com/var/www/subdomain/ .. I have to execute again: /var/www/subdomaincom/bin/console config system url https://sub.domain.com .. the console prints (as always): system.url <= https://inversion.tupambae.com and I can use the site again for several minutes before it falls back to the wrong behavior.
After a break I started to check out everything again and did several tests. I changed the order of the lines in local.config.php and managed to get the exact same lines and order in the bin/console prints of both sites. The problem stayed. Than I checked example_local.config.php to compare and found that actually any of the lines figure there, so I deleted them all together. Right now local.config.php looks like this:
The site still behaves the same way. At first it loads correctly, and quite fast, after some minutes it starts taking lot's of time and ends up like in the screen of the initial post.
Now the bin/console config print shows only the following two specific lines related to the whole path issue, all the other references mentioned above are gone. I have no idea how to delete the remaining references. Those lines have to figure somewhere in the DB I guess, but again, no idea what lines to execute to display them or delete them: config.system.basepath => / system.basepath => /var/www/subdomain
@utopiArte Once you set a config value using the console, changing the value in the config file won't help as the console stores the value in the DB and it has precedence over the files.
No matter how you did it, the important part is that system.basepath must be set to the install folder, and system.url must be set to the full URL. You can verify what the current value used by Friendica for either config keys is using the console.
I've browsed https://inversion.tupambae.com and when the style is missing, the first line shows inversion.tupambae.com/var/www/inversioncom even though it should show https://inversion.tupambae.com. This is governed by the system.url config key.
To me it "feels" that /var/www/subdomaincom is added by some mecanism to sub.domain.com. In particular because when setting specifically with the console command /var/www/subdomaincom/bin/console config system url -> https://sub.domain.com it works for some minutes.
As last intent I tried to use the following commands to reset the two mentioned settings but the behaviour ist still the same: /var/www/subdomaincom/bin/console config system url https://sub.domaincom /var/www/subdomaincom/bin/console config system basepath /var/www/subdomaincom/
No I can't. There is no change at all in the results of the bin/console config output (print). Actually even the settings send and "officially" accepted by the system do not show up in the print.
Latest thing I tried, totally feed up, was to send the commands several times and also making sure they end with a slash.
So I send: /var/www/subdomaincom/bin/console config system url https://sub.domain.com/ system.url <= inversion.tupambae.com/ /var/www/subdomaincom/bin/console config system basepath /var/www/subdomaincom/ system.basepath <= /var/www/inversioncom/ /var/www/subdomaincom/bin/console config system url https://sub.domain.com/ system.url <= inversion.tupambae.com/ /var/www/subdomaincom/bin/console config system basepath /var/www/subdomaincom/ system.basepath <= /var/www/inversioncom/
And right now it looks like it sticked, for now the site remains operative! 🙄
(please have in mind that this is 2021.04, if it weren't so outdated I'd say we need to figure out the problem because this feels like a real problem under the hood)
@utopiArte Oh I forgot this was such an old version. There still is an issue with the pictures because /var/www/subdomain/ gets prepended in the URL. Do you mind running bin/console config system and send the output to hypolite@mrpetovan.com ?
@utopiArte There's no file named boot.php at the root of the Friendica install, not sure where it's coming from.
The CURLE_OPERATION_TIMEOUTED constant is defined by the PHP curl extension, please verify it is enabled in the CLI version of php with php -i | grep curl.
For what it's worth, I found friendica to be unfriendly to non-root URLs, that is to say URLs that aren't at /, so I setup a subdomain specifically for friendica, friendica.eskimo.com/. You may wish to give that a try. The machine domain does not need to be this, you can do it by configuring DNS to point to the base IP of the machine and then configuring a subdomain in Apache or Nginx as a virtual host.
Also, this is a btop from an 18 core machine with 256GB of RAM with the major load being friendica, a secondary load is hubzilla, mastodon, and miskey, but friendica represents more load than the latter three combined, but it also gets more traffic than the latter three combined. But what I am getting at here, you may want to consider dedicating a physical server. I'll grant you it's busier than usual as we've just returned the server to service after three months of downtime, 7 motherbords, three power supplies, and two CPU's later.
Not mentioned in the inicial post or answers as of now is the setting of /sites-available/sub.domain.com-le-ssl.conf
sub.domain.com-le-ssl.conf <IfModule mod_ssl.c> <VirtualHost *:443> ServerAdmin admin@domain.com ServerName sub.domain.com DocumentRoot /var/www/subdomaincom <Directory /var/www/subdomaincom> Options Indexes FollowSymLinks MultiViews AllowOverride All Order allow,deny allow from all </Directory> <FilesMatch \.php$> # 2.4.10+ can proxy to unix socket SetHandler "proxy:unix:/run/php/phpXy4-fpm.sock|fcgi://localhost" </FilesMatch> # DirectoryIndex index.php ErrorLog ${APACHE_LOG_DIR}/subdomaincom_error.log CustomLog ${APACHE_LOG_DIR}/subdomaincom_access.log combined # RewriteEngine on # Some rewrite rules in this file were disabled on your HTTPS site, # because they have the potential to create redirection loops. # RewriteCond %{SERVER_NAME} =sub.domain.com # RewriteRule ^ https://%{SERVER_NAME}%{REQUEST_URI} [END,NE,R=permanent] Include /etc/letsencrypt/options-ssl-apache.conf SSLCertificateFile /etc/letsencrypt/live/subX.domain.com/fullchain.pem SSLCertificateKeyFile /etc/letsencrypt/live/subX.domain.com/privkey.pem </VirtualHost> </IfModule>
* Inure App Manager (Trial): multi-purpose app manager for all rooted and non-rooted devices 🛡️ * Dharmik Online: Browse Bhagavad Gita 🛡️
Dharmik Only replaces the Dharmik app at IzzyOnDroid. The latter includes all media and is fully offline, but unfortunately jumped to an APK size of > 300 MB…
This is a repository of apps to be used with your F-Droid client. Applications in this repository are official binaries built by the original application developers, taken from their resp. repositories (mostly Github, GitLab, Codeberg).
Chromium now has initial, experimental support for the xdg-session-management #wayland protocol, which will start shipping in canary channel in the coming days. I've implemented and tested it against Mutter 48, the only compositor supporting it atm - also experimentally - since version 47.
Short demo of Chromium M135 with Ozone/Wayland backend, showcasing its WIP support for the xdg-session-management-v1 protocol. Running unders a Gnome Shell 4...
Completely rewriting a multi-million line COBOL codebase that has life-or-death consequences for real people in the space of a few months, using gen AI?
I’ve been writing software for 40-some years, and I have to say: this may be, without exaggeration, the stupidest software-related idea I’ve ever heard from leadership.
#Mapycz (now #Mapycom) from #Seznam use #OpenStreetMap everywhere except their home country - Czech Republic, where they have their own proprietary maps. There have been rumours in the #OSM community that they actually copy from OSM. Today I found proof that they really do. I mapped a new park in #Brno just by surveying and taking notes last year. Later the park showed up on Mapy.com with exactly the same path layout. The new orthophoto is out and it revealed that I didn't get everything right and Seznam got it wrong too. Exactly the same mistakes.
Yes, the same happened to me a few years ago. I've added a house that wasn't in photo maps yet into OSM, and it did appear in Seznam Maps in the same exact shape.
@jmlich no, they are not saying that they use data about communications from OSM in the Czech Republic. You can't just blend OSM objects with proprietary objects of the same type. It's a license violation.
@jmlich Pepo, if you want to make judgments, please at least study the OSM license first. You cannot use OSM objects in a map where you already have proprietary objects of the same type. And you can list sources however you want. They use OSM for everything in other countries. That's fine. They only use OSM for certain types in CZ, such as benches, recycling points. That's fine. They use their own proprietary data for roads and paths and they can't add roads and paths from OSM to them.
additionally it is question if it is their fault. Maybe some 3rd party daty provider made shortcut by using OSM data without considering impact to licence. Anyway, you can ask them to fix attribution ..
@jmlich it isn't about fixing attribution. It's a license violation. It's like as if Microsoft took GPL code from Linux and put it in Windows without releasing Windows as GPL. Attribution doesn't fix it. Whether it was added by them or their supplier, is their problem and their legal responsibility. It has already been raised with Seznam BTW.
omg we should create librestreetmap, something with bsd license instead of gpl ;-D this rule that it cannot mix same types of objects in maps is terribily complex to explain. Or even that you cannot blend two images is something is irracional. At least for me. But IANAL of course
@jmlich yes, it's a bit complicated, but Seznam navigates well in it. All they are officially doing is in line with the license. Except for these "quiet copies". That's why I think it's not systematic, but rather a failure of an individual or individuals who wanted to make their job easier. For the sake of Seznam I hope they will stop it. We also wouldn't want an engineer in our kernel team who would copy code from BSD to Linux. 🙂
@organicmaps@mastodon.social this block should have been done earlier. It does have contributors from moscow, still spreading lies in Europe. Organic maps is a descendant of mail.ru's crap.me - NOT to be trusted!
@systemz @organicmaps I still hope it's just a failure of an individual or individuals who wanted to make their job easier and the company will fix it. It's only a problem for the Czech Republic, elsewhere it's pure OSM with their really good renderer on the top. I use OsmAnd myself.
@GuloGulo Ne, Locus Map nedělají to samé. Ti používají OSM mapy a ty vykreslují vlastním rendererem, to je naprosto v pořádku. Seznam v ČR používá své vlastní uzavřené mapy, do kterých zjevně kopíruje prvky z OSM, což je v rozporu s licencí. Není to žádná averze. Jen apel na Seznam, aby používal OSM v souladu s licencí. Jim by se taky nelíbilo, kdybychom data z jejich proprietárních map, za které třeba i zaplatili, kopírovali do OSM.
@GuloGulo Doporučuji nastudovat licenci OSM, než začnete vynášet takové soudy. Licence OSM umožňuje vzít objekty stejného typu a umístit je do vlastní mapy, ale nelze je míchat s objekty stejného typu pod jinou licencí. Seznam třeba bere z OSM lavičky, recyklační místa apod, ale nemíchá je s ničím jiným. Má tyto objekty pouze z OSM. V tomto případě bere z OSM objekty komunikací a přidává je do vlastních map, kde má komunikace pod proprietární licencí. To je porušení licence OSM. Locus Map nic takového nedělá.
Old thread but still. I thought this was a common practice - I'm from Slovakia and I remember I added some shops in #OSM and then when I checked #MapyCom , at least one of them was already in it with a slightly different name - probably added by user of #MapyCom . But then after the synchronization with #OSM (which happens every few days), they took just the shop I added (maybe merged them?). I suppose someone is sitting there and solves such conflicts
@efreet The thing is that unlike in Slovakia Seznam doesn't use OSM in Czechia. Except for a few object types: waste bins, benches... But definitely not roads or POIs which are part of their proprietary maps.
I wrote about how the user experience of encrypted messaging tools is sometimes just as important as the underlying tech. PGP, Jabber etc were not accessible. Signal makes it seamless. But the UI still has issues that can lead to mistakes 404media.co/behind-the-blog-fo…
For good reasons, people on Signal don't really want a public identifier, especially a phone number. (FWIW other more exotic apps like Briar and I think SimpleX [and GAcc-disconnected RCS] require you to set your own nickname for the contact.)
Elbows Up 🇨🇦 This morning, there are no fewer than four ANTI-Carney articles on the front page of the American-owned National Post. As expected, media is working hard to install a Trumpian hellscape in #Canada. Don't let them.
I saw Bloodywood last night. Pyrotechnics, headbanging, rapping/screaming, chugging on an 8-string guitar, and... thanking every single member of their (large) crew by name, with rounds of applause for each group, from their manager through to the tour bus driver. Lovely stuff.
Not sure what to think about this article on political polarization and extolling the virtues of moderation: onlys.ky/severance-and-the-div… particularly the calls to action (search for "3-3-3"). In particular, "Commit to engaging with three news sources outside your political comfort zone each week" seems like it would just be opening myself up to being taken in by propaganda. But then, am I too mentally weak to simply hear other ideas and opinions? I'd like to think I'm not.
I think consuming propaganda news sources at all is a bad idea. The idea that you can get at the truth by averaging lies is IMHO not valid. Engaging on a human level is a good idea though, and I could do better at that - I don't have any right wing friends.
Maybe reading also news from countries that do not have 2-party system could be better approach than reading the 'other side'. That is, if you want to read more news. Which I do less these days so, unfortunately, I can't recommend anything specific.
I think it depends a lot on the news source. I like reading/listening to the Bulwark. I don't always agree with them, but they usually get that most issues are complex/nuanced, and I don't want to be in an echo-chamber. Networks like Fox are mostly propaganda and not worth listening to imo, except maybe to understand what MAGA people are hearing (Decoding Fox News is good for that). And the center-right seems closer to the center-left than to the far right.
Recette pour une catastrophe : Prendre les 60 millions de lignes de COBOL qui font tourner le système informatique de la sécurité sociale et faire tout ré-écrire en Java par une IA. POPCORN wired.com/story/doge-rebuild-s…
(ça pourrait presque être drôle si ce n'était pas encore les pauvres qui allaient trinquer.)
I haven't started watching this series yet but I will soon, I think, and the score of movies and shows often really sticks with me. Might grab these preemptively, while carefully not looking at the track names.
🔧 Z Oscloudu zas vytahuju jednu z těch nenápadných, ale užitečných aplikací – SearxNG.
🔍 https://searxng.cz
Soukromý metavyhledávač bez reklam, profilování a šmírování.
Spolehlivě běží už pěkně dlouho a pomáhá s hledáním každý den.
Stop talking about AI already. Marvel at 249 pages of #Ghibli #GIFs posted on this ancient Tumblr from the internet's golden age. Every two-second GIF is bursting with life.
Because of corporate collusion with law enforcement dating back to the 80s, all colour printers put tracking information on their printouts, whether you're printing a colour document or not, encoding information that ties printouts back to the specific printer used.
Relatedly: don't forget that office and networked printers aren't really "printers" and haven't been for decades; they're networked servers that happen to output printed media. They store userID and datestamped logs of everything anyone prints on them, and nobody throws out logs.
Čas vyzkoušet nové prostředí – #GNOME48 na #ArchLinux 🎉 #i3 zůstává mým hlavním prostředím pro každodenní práci, ale musím říct, že GNOME 48 vypadá skvěle.
Final update: The developer is now on Mastodon via @andrew_guide.
Update: The developer has removed the ability to download Guide until the security issues mentioned in the linked thread are fixed.
Update: this product contains some code flaws that are concerning from a security perspective, beyond just giving control of your computer to an LLM. You might want to read this thread before installing the product: toot.cafe/@matt/114258349401221651
Update: I've exchanged some long emails with Andrew, the lead developer. He's open to dialogue, and moving the project in the right direction: well-scoped single tasks, more granular controls and permissions, etc. He doesn't strike me as an #AI maximalist can and should do everything all the time kind of guy. He's also investigating deeper screen reader interaction, to let AI just do the things we can't do that it's best at. I stand by my thoughts that the project isn't yet ready for prime time. But as someone else in the thread said, I don't think it should be written off entirely as yet another "AI will save us from inaccessibility" hype train. There is, in fact, something here if it gets polished and scoped a bit more.
Just tried guide for fun. It's supposed to be an app to use #AI to help #blind folks get things done. I asked "Where are the best liver and onions in Ottawa?" It: 1. Decided it needed to search the web. 2. Thought that the "stardew access" icon on my desktop was a kind of web browser, so clicked it. 3. Imagined an "accept cookies" dialogue it needed to accept. 4. Decided that didn't work, so looked for Google Chrome (I don't have chrome installed on that machine) 5. Finally opened edge from the start menu. By the way, it just...left Stardew open and running. Because apparently having Stardew Valley running in the background is a vital part of finding liver and onions in Ottawa. 6. Opened a random extension from my edge toolbar (goodlinks). 7. Clicked the address bar and loaded google.com, instead of just doing the search right from the address bar. 8. Got blocked because it couldn't sign into my Google account, even though it could have also searched from the Google homepage.
To be fair to AI, that was the kind of open-ended task AI is terrible at. If I had asked it to check an inaccessible checkbox, or read a screenshot, or something, I'm sure it would have been fine.
Anyway, I'm still better at using a computer than an AI. So is my 87 year old grandfather, for that matter. www.guideinteraction.com
With Guide, inaccessible doesn't have to mean undoable. Guide is a Windows AI assistant that helps people with low vision or blindness navigate the digital world.
There's a new product that has been gaining some buzz in the blind community, a Windows app called Guide that uses AI to perform tasks on your computer. It's pitched as a way to get around web accessibility problems in particular. I won't link to the thing itself, because I don't want to give it that validation, but I'll link to a previous discussion thread about it: fed.interfree.ca/notes/a5wf4ys…
I've spent some time taking this app apart. The level of shoddy work here is deeply disgusting. 1/?
So I wanted to try guide on a real accessibility issue. However, it seems that #codeberg has finally fixed their #inaccessible#captcha. Now, if you tab into the #captcha field, you're told what you need to type to get past it. Good job codeberg! #a11y
is codeberg staying online consistently for you? Last I heard, they were still combatting AI training models, or something AI-related, that was sapping up all of their resources.
Made me revisit how things in Forgejo were progressing toward federating pull requests without resorting to emailing patches.
I've had an account to username squat mainly, and had lofty goals of migrating off of GitHub, but hadn't put in the time to test the CICD system. Didn't want to tax the community resource if BigCorp was willing to foot the bill elsewhere. Then again, Copilot training. Ugh.
This seems vastly overkill for what I want, a captcha solver, but do you have any suggestions on how to make this solve image only captchas? Or if there's a better tool just for that task I'd love it!
I guess you didn't notice me using the AI feature in the be My Eyes app for Windows to read handwritten scanned PDF's, but putting aside your attempt at snark, would love suggestions in this area. You can criticize tech culture and still use tech, you know.
@WeirdWriter Your criticism is most likely to harm the access that even you apparently hypocritically rely on. So I’m unlikely to recommend you something I rely on for my living, lest you insight a backlash against it. It’s a matter of trust.
@WeirdWriter I don’t post about the AI that actually works on the fediverse. The anti AI crowd here is likely to get upset that it can solve captchas, describe adult images, or make infographics for me, and attempt to shut it down or take it away.
I can assure you none of us have that kind of power. If anything, it will be a lack of long term financial investments for these companies that causes them to go under, which is outside the critiques of the Fediverse.
@WeirdWriter Maybe not. But I’m not suffering under the delusion that the Luddite’s and anti AI crowd are my friends anymore than the gross capitalist growth at all costs tech bros are. As blind folks we have to take what we can from both. Because both will screw us if it gets them an inch closer to their goals. So I keep the AI that actually works off the fediverse.
Shame because the luddites are making some pretty cool scripts and software. But while I believe the worst that can happen is you'll just get scolded by anti-AI folks, I actually fully support doing what you gotta do to survive under this world.
Anyone who is anti AI has never actually used it properly, they probably used it ones, it’s made a mistake or hallucination, and then they think it’s total rubbish.
@oasisfan@WeirdWriter I wouldn't go that far. But most anti AI folks are fanatics, and thus dangerous to accessibility, as they will happily throw the baby out with the bathwater.
First, it's an Electron+Python monstrosity. Specifically, the Python backend runs as a web server on the local machine, and the Electron frontend connects to that local web server. Along with the size of Electron itself, the frontend app is about 27 MB, mostly a node_modules tree with no hint of tree-shaking / dead code elimination. The front-end JavaScript code is not minified at all, so once you extract the .asar file, it's easy to look at it. 2/?
Yep, it can solve those CAPTCHAs where you have to choose the right picture, but the cloud vision NVDA add-on can do this, it just won’t click on things for you, but it can tell you what the pictures are.
@oasisfan@WeirdWriter I don't trust cloud vision. It's an addon written in Russian, that sends all data to a Russian domain. Including your Be My Eyes username and password if you give it that. If it was communicating with the services directly, fine. But if you check the sourcecode, it's not.
I think the lack of a privacy/"what I'll do with your data" section of the website is a big miss for a product that will have access to do whatever it feels like on my computer. @fastfinge
@jscholes@Jage On a more serious note, I think the interface as presented is just way, way too generalist, and its freedom too unrestricted. Things I'd like to see: 1. it gets prompted with the name of the currently focused app, and the mouse is disallowed from leaving that app window. 2. Bringing it up with alt+ctrl+g gives prompts of what tasks it thinks it could perform inside the current app. Instead of the current general "ask me to do anything!" 3. It should have access to the DOM for browsers so it's not taking screenshots and acting only on that all the time if it's being asked to do something on a webpage. 4. It really, really needs training on NVDA focus mode. It says to turn off speech, but that doesn't solve for it trying to type into an edit field when NVDA isn't in focus mode. It does this constantly.
I still don't think it'd be ready for prime time, but it would be closer.
@jscholes@Jage It's also a security thing. Sure, it was pretty funny that it opened Stardoo Valley. But if I was using it on my real machine, I can think of a lot of apps I wouldn't want it to unexpectedly launch.
@starkraving666 It's useful for a lot of things. Narrowly defined tasks like "describe this image" or "What is this control" or "what does this say" especially. But we're trying to create giant mega-models with all the data instead of tiny specialized models with the right data. I don't need the same AI that gives me alt-text for images I can't see to also write me python code and draw me a unicorn and tell me a story. If I want those things (and sometimes I do), they should be different specialized AI's that just do that one thing. It's okay if the Python-writing AI can't read me the menu when I go out to eat.
this kind of thing, where you just give an LLM full access to fuck around on your computer, terrifies me. Even more so that it's designed to be used by people who literally can't see what it's doing. How can anyone think this is safe?
We can hear exactly what it’s doing with our screen readers. And there is a stop hot key. I’m not concerned about safety. I’m more concerned about privacy, and correctly setting expectations. I work in tech, and I spend at least a couple of hours a day struggling with inaccessible apps and websites, and have to have a co-worker do certain tasks for me. Anything that can make this better is of interest to me. Sure, everyone should build accessible apps. But for the most part, nobody bothers. But I still have to work. However, solutions like this need to be strictly sandboxed to the app I actually need help with, not given access to the entire computer. Not only is this more secure, private, and safer, it also works better. The AI doesn’t need to get distracted by the fact that I have BBalatro installed, or that a windows notification popped up. Just give it the current window.Also, the examples for use need to be more like “Ask me to check a keyboard inaccessible checkbox for you” or “I can help you read an image without alt text”. The current examples are more like “Ask me to book a flight for you”. In its current form, AI can’t do that. The developer has strongly committed to fixing both of these things.
muž s klapkami na očích
in reply to Tu nube seca mi río • • •