Politici o problémech pošty věděli. Neudělali ale nic
neovlivni.cz/politici-o-proble…

Oh, #screenreaders cutting off long #alttext is a myth? yatil.net/blog/there-is-no-cha…

That is fantastic news, I was really worried we'd have to do ridiculous things with longdesc (deprecated) or visible alt text.

Ahh! Finally.

in reply to Helena 🏳️‍⚧️

I've never heard of the 125 character limit? Maybe it was with screen readers other than NVDA? I thoguht the limitation was from the application side. Twitter & LinkedIn both allow 1,000 characters. I checked for someone last year & Facebook allowed me to post the entire Wikipedia WWI page as alt text. NVDA read through the first 42,158 of 259,432 characters. Not sure if that limitation is from facebook's side or NVDA's TBH, but I don't anticipate many real world users hitting it
in reply to NV Access

@NVAccess if you haven't heard of it, then we've been reading different articles 😅
css-tricks.com/just-how-long-s… indeed a search for "screenreader alt text 125 characters" finds a plethora of articles probably all copying from each other.

And I guess from this myth, lots of applications set limits which have changed a lot over time.

Great to have another data point that it works for NVDA :) It works fine for me with TalkBack as well.

in reply to Helena 🏳️‍⚧️

Someone asked me about limits last year, and that must be where they got the idea from. FWIW, at the time (June last year), LinkedIn only allowed 300 chars where it now allows 1,000. I haven't tried dropping an entire Wikipedia page in Facebook's alt text today, but based on my result from last year, if there is a limit there (lower than 42k), it will be on the Facbeook side.
in reply to Helena 🏳️‍⚧️

@NVAccess and re: your mastodon comment, yeah, every mastodon server can set different limits. Just like every other website, hence the question for me as a website operator: do I set a limit.

The specific use case for me is online scientific tutorial content, where the images can be quite complex

So, having real hard evidence that there is no limit is extremely useful to know and I can stop thinking about how best to support long descriptions.

in reply to Helena 🏳️‍⚧️

I guess the question is how will having (or not) a limit affect your users? It may encourage people to be more succint - and alt text is supposed to be succinct. But whatever limit you set, you will then potentially have people say "But I need more room to describe this really complex chart!". I haven't seen many complaints about limits of around a thousand characters. It's double the average Mastodon post lengh & 4 times a tweet so that might be a place to start if you want a limit.
in reply to NV Access

@NVAccess yeah given what you and others have said, we will not impose a limit.

Conciseness is good, but getting scientists to write ANY alt text is not a fun experience, so, it is unlikely they would even reach 1k characters.

It was /only/ going to be necessary if some screenreaders would just stop announcing, then we'd probably still support longer but have to write some complicated workarounds to ensure that it got chopped up into smaller blocks when too long, or announced as "follow the next link for the long textual description of this image" which would let the user navigate to a text block and then somehow back.

That was a development activity I was not especially looking forward, and now I can just mark the issue resolved and move on! :ms_happy: Less work for me, I'm happy.

in reply to Helena 🏳️‍⚧️

& from our side - getting some web developers to write even vaguely accessible code sounds like a similar exercise to getting your scientists to write alt text (big generalisation I know!) - so for someone like you who IS doing the right thing, the last thing we want to do is make you do MORE work. If there was a limit from the screen-reader side, then I'd definitely encourage you to contact us (if it's in NVDA at least) as that is something we should be looking to fix rather than you.

Do you know there's a #fediverse alternative to Amazon-owned #Goodreads? #BookWyrm is a social network for tracking your reading, writing reviews, and discovering what to read next. You can follow and interact with users on different #BookWyrm instances and on #Mastodon. You can import from a Goodreads CSV export. You can create private shelves and curated lists. Join us at ramblingreaders.org or choose one of the other instances available #books #reading #bookstodon @bookstodon

Release time! 🎉 Our team has been hard at work to push an update to Tutanota. 🥳 You can now

* Create repeating events & delete individual entries

* Enjoy more space for reading emails on mobile clients as we've moved the subject line into the header

github.com/tutao/tutanota/rele…
--
Unser Team hat hart daran gearbeitet, ein Update für Tutanota zu veröffentlichen. 🥳 Du kannst jetzt

* Wiederkehrende Termine erstellen & einzelne Einträge löschen

* Mehr Platz zum Lesen von E-Mails auf mobilen Clients

A very useful article. Method 2 was the best one, but you apply it at your own risk. How to Restore Old Classic Notepad in Windows 11. winhelponline.com/blog/restore…

Pitermach reshared this.

Tutanota legt großen Wert auf Sicherheit und Privatsphäre. Dazu gehört auch, dass man sich anonym anmelden kann - ohne Telefonnummer!

Es gibt nicht viele, die sich so klar auf Privatsphäre und Anonymität fokussieren. Check out:

@threemaapp

@torproject

@Tutanota

tutanota.com/blog/posts/anonym…

in reply to Tuta

Was ich aber nicht verstehe:

Warum wird Threema als Privatspärenfreundlich und Anonym dargestellt wenn es explizit nur per App auf einem Telefon funktionieren will.

Für ein Mobilfon müßte ich doch erstmal sprichwörtlich die Hosen runterlassen und mich mit verfifizierter Addresse und Videochat ausweisen damit ich eine SIM-Karte bekomme.

Von daher kann ich Matrix und XMPP sehr viel mehr empfehlen

Oh, and btw, a tiny little #AndroidAppRain also at apt.izzysoft.de/fdroid with 6 updated and 1 new apps:

* JRPN 15C: RPN Calculator inspired by HP 15C Scientific Calculator

Enjoy your #free #Android #apps with #FDroid and the #IzzySoftRepo :awesome:

Tutanota focuses on security and privacy. Part of this is an anonymous way to sign up - without a phone number!

There are not many that focus so clearly on privacy and anonymity. Check out:

@threemaapp

@torproject

@Tutanota

tutanota.com/blog/posts/anonym…

Sur pablockchain.fr, retrouvez mes articles 📰, billets de blog 📝, et conférences 👨‍🏫 à propos des #blockchains, des #cryptomonnaies, des #NFT, du #web3, etc.

📮 Vous pouvez aussi m'y laisser votre adresse email pour être informé·e de mes futures interventions 😊.

Čech na internetu sdílel tisíc písniček. Policie ho zadržela a teď mu hrozí dva roky vězení ift.tt/rec4kAb

TL;DR #HP printers and ink cartridges have #RegionLocks, just like DVDs. Buy a brother laser printer, folks. theverge.com/23648726/hp-offic…

Region-locked globalisation. Fun! … not.

This entry was edited (2 years ago)

We're back from #GDC23, where @godotengine had an official booth for the first time! :godot:

It was an amazing experience, with 16+ indie games presented by their developers, tons of great discussions with visitors, talks, and an IGF award! 🏅

Our retrospective:
youtube.com/watch?v=vdAwlI6NG0…

This entry was edited (2 years ago)
in reply to Godot Engine

We had our own Godot GDC Meetup hosted by @GitHub, which was a great time to meet old and new friends in a relaxed environment.

@gdquest gave an insightful talk on the governance challenges faced by big #opensource projects like Godot:
youtube.com/watch?v=ZRFv2rTJLU…

in reply to Godot Engine

And @reduz gave an inspirational talk about #GodotEngine's recent growth, and its development into an open ecosystem, similar to other big successful open source projects like Linux.
youtube.com/watch?v=XduuHNOGGq…

Since nightly-2023-04-01, this Rust code can panic:

unsafe {*(1 as *const i32)}

because rustc now inserts a check that raw pointers are aligned before dereferencing them.

github.com/rust-lang/rust/pull…

This check is only inserted when debug assertions are enabled. If you have any notable experience with this, drop me a line!

"Cat pictures and videos are often shared across the Internet. Many of these files display feline subjects interacting with boxes and other containers.

Since there is currently no compact notation for describing such media, this document details a standard notation to describe the position and interaction of cats, containers, and related subjects pictured in these images."

RFC 9402: Concat Notation
rfc-editor.org/info/rfc9402

This entry was edited (2 years ago)

Research Article
Knitted origami
Elizabeth L. Wilmer

"Techniques are presented for embedding horizontal, vertical, and45◦diagonal crease lines into garter stitch knitted fabric. While theseare mostly based on standard lace knitting stitches, the horizon-tal creases use double-cable-crossed elongated stitches in a non-standard way. This crease library suffices to knit a model of a squaretwist, a foundational origami tessellation unit."

tandfonline.com/doi/epdf/10.10…

I think centralized, large predictive text models (no, they’re not AI—just ask Qui-Gon Jinn!) are missing the point. The more interesting application than literally mindless spouting and regurgitating things that have been said on the internet is specialized, localized models trained on very specific data.

But most importantly, can people please stop pretending text models are somehow inherently intelligent?