7/20
#20albums20days
Jak je v GrapheneOs vše v sandboxu, musím povolit v /Nastavení/Sandboxed Google Play oprávnění 🤦🏽♂️🤦🏽♂️
Teď to jdu vyzkoušet, myslím, že to půjde.
grapheneos.org/usage#android-a…
Wait, what is this?
An HTML message with hardcoded white background and dark text that works perfectly out of the box in dark mode in Thunderbird without any add-on?
Yup!
Coming soon on Daily!
As someone who has to write emails that work in different clients: oh no 😅
Edit: which is not to say that I don't think users will appreciate it :)
Random thoughts.
So broadly speaking we can look at the defense-in-depth against spam-like attacks as happening in a few different layers.
Automatic:
1. Front-line preventative. How do we prevent a server compromise in the first place? This is your entry friction.
2. Spread preventative. Given that a server is compromised, how do we prevent it from arriving on other servers?
3. Local visibility limitation. Given that the message is now present, how do we prevent it from being seen?
1/
Huh. I'm monitoring stack consumption in librsvg, and (kinda of course) it consumes much less stack when built in release mode than in debug mode.
But... I have this function with a huge closure, and that closure eats 7 KB of stack in debug mode... and 0 KB in release mode (!)
So, yeah, lots of local variables (it's a huge ugly closure), but goddamn.
We had the remarkably effective trick of "throw Alex at the problem, he'll probably figure it out" 😅 github.com/bytecodealliance/wa…
You could see whether doc.rust-lang.org/beta/unstabl… helps you at all, and perhaps the cargo-call-stack tool that's built on it, but the list of known issues to both of those is quite long.
When I've needed to understand stack usage in code generated by Wasmtime, I've looked at the disassembly for the specific prologue instructions that allocate the stack frame. I haven't tried doing that for LLVM-generated code and have no idea if it would be helpful.
In short, the answer to your question is "no", I don't have good tricks for this 😅
Motto for my #musl/software ecosystems work: Move slow and fix things.
Motto for my #3dprinting work: Move fast without breaking things.
Disabled people only!
Do you consider person-first language to be ableist, vs. identity-first language?
My opinion is in the next post!
🔁 Boost for reach
- Yes (37%, 20 votes)
- No (26%, 14 votes)
- I don't know what that is (17%, 9 votes)
- I'm not disabled but I like polls (18%, 10 votes)
A Friday news dump from yours truly: I called up Be My Eyes CEO Mike Buckley this morning to pick his brain about the recently-announced collab with Meta and the Ray-Ban smart glasses.
Amongst other things, Buckley said Meta CTO Andrew Bosworth was key to the whole thing.
Apple did the research; LLMs cannot do formal reasoning. Results change by as much as 10% if something as basic as the names change.
reshared this
If you ever doubt that billionaires are completely alien compared to the other 99.9999% of us, remember that out of the ~400 tech billionaires out there, not a single one has stepped up with their pocket change to make a life-changing donation to the Internet Archive or the now defunct Living Computers Museum.
Out of 400 supposed techies, not a single one has an interest in data preservation or retro computing??
Clearly they have no passion for technology (if anything) beyond a means of further wealth accumulation, having more in common with the dragons of yore, sitting upon a hoard of gold and demanding further human sacrifice while giving nothing in return.
reshared this
@matt
no, he had a personal interest in the computers he used in his youth, built a playpen for himself, then let it be destroyed when he died, following the billionaire philosophy of "me me me"
he had years to actually transfer the computers to the museum instead of them being on loan, and never did
Delivering the final talk at the last ever XOXO was a huge responsibility, a tremendous honor, and an absolute joy.
I'm so stoked that now you can watch the video of me explaining why nothing lasts, and why this gives me hope we can create moments that'll help change the world and save each other.
OK, es ist (fast) Wochenende und wieder Zeit für ein Experiment – diesmal zum Mitmachen 😉:
Welcher der beiden nachfolgenden Posts sieht für euch besser aus und auf welchen würdet ihr eher klicken.
Gerne schön weit teilen, damit das von vielen und mit den unterschiedlichsten Clients gesehen wird.
Wenn der Unterschied besonders groß ist, gerne auch mit Screenshots antworten. Danke!
#Mastodon #SocialMedia #Fediverse #Apps
- Post 1 (mit generierter Preview) (75%, 57 votes)
- Post 2 (mit Titelbild als Anhang) (25%, 19 votes)
Die Darstellung hängt ja auch vom verwendeten Clienten ab...
Bei #FediLab hab ich z.B. die Preview ausgeschaltet und bei #Sharkey bekomme ich eine Vorschau auch erst, wenn ich das will, wenn ich mit der Maus oder dem Finger drüber bin.
Das angehängte Bild halte ich für überflüssig.
Sinnvoll fände ich, wenn Überschriften mit Markdown/ MFM als solche gekennzeichnet würden.
Mittels 𝙐𝙣𝙞𝙘𝙤𝙙𝙚 ist sowas möglich. Ein Konverter, wie anytexteditor.com/de/unicode-t… kann da hilfreich sein. 😁
#Barrierefreiheit #blind
So, the ham radio folks who have been telling me there will be no fallout from Israel using exploding (fake/counterfeit) Icoms
Flightaware: "Dubai's Emirates Airlines has banned passengers from carrying pagers and walkie-talkies on its flights, following last month's attacks on Lebanese group Hezbollah involving communication devices that exploded."
flightaware.com/squawks/view/1…
i still kind of love Fedi but i don’t enjoy loving it.
most people still don’t see any nuance beyond sides in a conflict (I cannot emphasize this enough), and people offering solutions tend to either be too far from these conflicts or too close to them and it’s not always their fault.
the network is incredibly fragile. it’s laughably easy to bring an instance to its knees, especially servers like [redacted].
good-vs-evil thinking beyond extremely obvious cases makes people, including past versions of myself, extremely easy to manipulate at scale. As Fedi is much more tribal than most other forms of social media, this naturally makes it far too easy to stir up anything from drama to mass calls to action based on flimsy selective reasoning.
And yet, it’s the only network with the people I want to follow. And the only place that I’m reasonably confident will still exist in the future. And that’s before you account for the sunk-cost fallacy impairing my judgement.
Seirdy reshared this.
one of several problems with whittling down any conflict to “sides” is that you eventually find yourself on certain sides and that becomes part of your identity. And when someone on one of your sides does something horrible that hurts another one of your sides, you end up having an identity crisis and need to redefine what those sides are to exclude that person.
Hence: whitewashing in response to a BIPoC user opposing you, erasing queer identities or redefining queerness to maintain purity and exclude certain queer people.
Previously on “Seirdy posts about meta non-specifically”:
this was prompted not by big new meta but some introspection as I danced on the event horizon of something resembling burnout (I find “burnout” too binary/absolute a term to describe how I run out of spoons).
Seirdy reshared this.
just re-read the post and I want to clarify: there are legitimate reasons to create new queer labels that are more restrictive if you want to describe an experience that’s unique to a smaller group of people or class. i’m not against that entire practice; using it to describe more specific experiences and exclude people who can’t relate to it is a good thing.
I’ve seen it used multiple times to redefine queerness on Fedi to exclude people on a queer instance who aren’t aligned with another queer instance, because the first instance isn’t ✨actually✨ queer, because queerness means agreeing with me! That’s the specific form I’m talking about.
I mean, nobody says it like that. People dress it up. But everybody knows the feelings behind the words.
this, despite the many (god, so many) pronunciation mistakes, is a masterpiece.
I recommend that all who consider themselves allies to Jews and Palestinians watch this video essay (or even better, read the sources that were used to write it. but watching it is shorter).
Nebula
nebula.tv/videos/jessiegender-…
YouTube
youtu.be/zOmvJZCn-IE
#Zionism #Antizionism #Israel #Palestine #Genocide #antisemitism #Gaza #VideoEssay
The latest edition of This Week in Matrix is out! Policy lists for the community bot, aaaaaallllll the spec versions for the Elm SDK and improvements on the Neoboard widget.
This and more on our blog at matrix.org/blog/2024/10/11/thi…
I’m not sure whether there is anything but SquareX PR to this story: globenewswire.com/news-release…
Either way: yes, malicious browser extensions can still do considerable damage, even with Manifest V3. No amount of technical improvements is going to change that: as long as extensions can do something useful, they will also be able to turn rogue.
Manifest V3 implemented some rather effective measures to prevent legitimate extensions from being exploited. It also to some degree made it easier to inspect extensions and identify problematic use of browser APIs.
But, and I doubt that it comes as much of a surprise, it’s still very bad to have malicious extensions installed. No real changes here.
Archos
in reply to Vojtěch Dubský • • •Štěpán Škorpil
in reply to Vojtěch Dubský • • •