fabrice.capyloon.org
Web geek
382 posts
238 followers
163 following
Getting Started
Active Commenter
comment in response to
post
I think you're correct about the use cases. Streaming support was also not that easy to spec at the time, because JS Streams were not a thing yet. Today we could hook crypto algorithm in a developer.mozilla.org/en-US/docs/W... !
comment in response to
post
Some algorithms, and the lack of proper streaming encryption/decryption was a blocker for media & files support.
comment in response to
post
I hope WebCrypto will get what's required for a MLS implementationi. When we did WhatsApp for KaiOS we could only implement 1-1 chats with WebCrypto, but not group chats.
comment in response to
post
You can give them money, that will add a badge to your profile pic
comment in response to
post
We absolutely should. Piggy back on what the Linux Mobile community is doing for the HW support, and slap a composable web OS on top. No concessions, no "we should support android apps" BS.
comment in response to
post
Probably only costs 1.5x browser funding!
comment in response to
post
Read-only access is not what I call a shared drive, you are running in easy mode :)
comment in response to
post
Unfortunately, Android has made changes which will make it much harder for us to port to Android 16 and future releases. It will also make adding support for new Pixels much more difficult. We're likely going to need to focus on making GrapheneOS devices sooner than we expected.
comment in response to
post
Your browser knowing everything about you is not the problem. The data being on TBC servers on the other hand...
comment in response to
post
And to no one's surprise: www.reddit.com/r/browsers/c...
comment in response to
post
As long as we don't throw away the unique differentiators of the web, anything should be fair game for change. Just doing a copycat of native platforms but with js/html/css instead of kotlin / swift is not gonna save the Web though.
comment in response to
post
99.99% of users don't ever configure their routers or anything like that.
comment in response to
post
I don't think regular users have any idea what you're talking about, and thus can't give informed consent.
comment in response to
post
I'm curious to see how the prompt will read. It's almost like we need an improved trust model for web apps.
comment in response to
post
blink twice if you need help
comment in response to
post
Not so much when the "funded by" starts arguing that the business model of the "made by" should be protected.
comment in response to
post
HAAAA www.lemonde.fr/politique/ar...
comment in response to
post
Peut-être... Tondelier & Ruffin ? Tant qu'ils ne passent pas 6 mois à se pourrir dans une primaire on pourra s'estimer heureux
comment in response to
post
You're not wrong but I don't see anyone with the required charisma on the French left for that. We need the equivalent of a Bernie & AOC duo. Combative but not whining.
comment in response to
post
Write a watchdog that can be fed with (pid, path_to_app) and will remove the app when the process dies. Surely someone can vibe code that in 2.34 seconds.
comment in response to
post
seize the means of operating your systems
comment in response to
post
That's their way to feel like C++ developers doing LTO builds
comment in response to
post
Douche = humidité, c'est pas l'idéal de combiner ça avec un penderie (potentiellement des soucis de moisissures dans les vêtements).
comment in response to
post
And did you really buy their sycophancy excuse? They will have to come with new rare words for any screw up instead of telling it like it is: our model is worse than the previous one
comment in response to
post
This is a very specific test: "This evaluates how often an LLM introduces hallucinations when summarizing a document", not representative of open questions.
I wish people learn some information theory to understand that you can't pack knowledge without loss forever.
comment in response to
post
Are you sure? Just last week OpenAI pulled out their newer model because it was hallucinating even more than the previous ones.
comment in response to
post
Interesting that LLM/AI proponents are pushing so hard on that energy topic recently, while at the same time ignoring ethical issues and the fact that models seem to reach a quality plateau. "We're destroying the web but not the planet!" is not a great flex.
comment in response to
post
If you look at what they are pushing for these days, this is: "anonymous" Ads, and "private" AI.
comment in response to
post
Oh they understand that. They just have no idea what to build (and also make dumb mistakes every other quarter).
comment in response to
post
If the token is shared state between tabs they should use the BroadcastChannel api to synchonize themselves.
comment in response to
post
They have a lot of cash to bootstrap that, and of course that would be a paid offering.
comment in response to
post
They could claim that their user base is not going down that fast: data.firefox.com/dashboard/us...
comment in response to
post
No... there are other options. Mozilla has a very good client side encrypted sync service, they could leverage it to eg. store all your files and build a whole ecosystem around that.
comment in response to
post
Booking flights + Hotel + Car has been the classic demo of recommendation systems since the early 2000s. It's just so dumb and boring... Most systems don't even do a better job than expedia/booking.com, and most people don't travel anyway
comment in response to
post
Add UCAN support to the PDS (or whatever new smart thing @expede.wtf is designing).
comment in response to
post
I fully agree but I've never seen a design that ensures that you can't scale (apart from blockchains, ha ha).
comment in response to
post
I feel that there's a (large? I'm not sure) part of the fediverse actors that actively rejects scaling - so they see usability friction as a feature, not a bug.
comment in response to
post
Was DID resolution also still working?
comment in response to
post
And the AI guy still can't code. But he has a new tool that let him build apps - these are two different things, and maybe it's ok?
A bit like some cooking robots make it possible for less skilled cooks to prepare some recipes.
comment in response to
post
I just have a very hard time calling "private" data that is in the clear on someone else's machine. Maybe I'm too paranoid, and for sure nothing prevents you from storing opaque encrypted blobs in a PDS.
comment in response to
post
But no-one ships stock Android/AOSP, so that comparison doesn't make any sense.
Instead of forking of Lineage, they could fork of Graphene and benefit from a much better base.
comment in response to
post
Ok, but I still think this changes the threat model and the kind of target a PDS can become.
comment in response to
post
What I like about current PDSes is that they are relatively dumb/simple data stores, clearly holding public data.
This kind of proposal increases complexity, and will make PDSes juicer targets for attackers because people will start to use them for private data but without encryption.