I like it, though there may be some bias involved... docs.dxos.org/echo/introdu...

I'm based nowhere near LA, but funnily enough I'm going to be in LA around this time. Will add this to my calendar and see if I'm able to make it work. Would be fun to meet some atproto folks irl!

Why do people on the internet like to assume the worst of others?

Of course there’s always still risk of factions becoming isolationist, but the more apps, connections and interop there is the more painful that becomes because they lose access to the greater ecosystem.

I guess what I’m saying is that you’re missing some of the incentives here. The more interop there is, it’s more incentive to keep the network open because the openness provides value.

This is why it’s so important and exciting that there’s so much activity in the atproto developer community right now. May a thousand flowers bloom! 🌺 🌼 🌷

I agree with Erlend that your threat modelling is off here. As interop increases, the value of connectivity increases and it makes severing from the network more painful. Even a selfish actor would see that there’s more value in keeping it open if locking it down would kill >50% of the functionality

Familiar with UCAN, that makes sense 👍 so would the flow look something like:
1. app loads tile manifest
2. app prompts user to grant permission to resources
3. if granted, app issues credential to the tile
4. tile uses the credential to authenticate the wish connection to data source

Wondering if this scenario fits with wishes or if its outside the scope:
1. wish for access to local db with permissions of A
2. host opens rpc connection with tile
3. tile has gated reactive data access to the subset of the db

Sorry about that, I’ve flagged it to Rich. If you’re looking to be in touch the best thing to do is dm richburdon on Discord.

For us, the AT integration has been a priorities question for a while. That’s probably still the case…but I will try to find some time to poke around web tiles

@dxos.org is all in on web tech and it’s felt clear to me for a while the AT would be a great fit for a plugin registry as well so it feel like it would align quite well.

I'm very interested in this personally but also think that @dxos.org has interest in something like this for Composer in the long run. As I've had time, I've been poking around the edges of what it would look like to have interoperable code between Composer & other apps like Patchwork.

I have a feeling the rise of Claude Code and personal software changes the dynamics here. Sure, maybe its less likely that a company will come in and build on top of some one else's format because they want to control their own schema, but if I can do that cheaply just for myself, I probably would.

GitHub - wittjosiah/npm-trusted-publisher: Tools to automate npm trusted publisher configuration for GitHub Actions OIDC Tools to automate npm trusted publisher configuration for GitHub Actions OIDC - wittjosiah/npm-trusted-publisher

It’s baffling to me that there’s no straightforward way to do this at the org level.

Anyways, I published the code here in case it helps anyone else in a similar situation: github.com/wittjosiah/n...

The crazy thing is this whole thing with Claude Code definitely took me less time than doing all the packages manually because I tried doing that and gave up after extrapolating how long it would take to get through all the packages.

At first I tried getting it to write playwright script to do it but it pretty quickly got blocked as a bot. Then I got it to generate a chrome extension to fill out every for me and auto-navigate through a list of packages. That worked, presumably because of more human in the loop.

NPM publishing is such a pain. I understand the reason behind their move with granular tokens but it was managed so poorly. This week finally got around to migrating all the DXOS packages to trusted publishing and was only able to do it because of Claude Code.

A wide, snow-covered outdoor plaza overlooking a frozen marina, with two tall decorative metal totem-like columns in the foreground. Steps and railings are partially buried in snow, benches dot the open space, and a flat winter landscape stretches to the horizon under a blue sky with scattered clouds.

A snowy urban sidewalk with several bike-share bicycles locked in a row of metal racks, partially buried in snow. A plowed road runs alongside the sidewalk, with a 30 km/h speed limit sign visible. Apartment buildings and leafless trees stand in the background as the low winter sun sets behind them.

A snow-covered sidewalk beside a plowed road, with a row of bike-share bicycles locked into metal racks. Bikes are dug out of the snow, and a snow shovel leans against it in the foreground. A “Maximum 30” speed limit sign stands along the road, with bare trees, street lamps, and apartment buildings visible under a winter sunset sky.

Snow day! ❄️

thanks, this is awesome! I'm glad to see that this sort of thing works

Block Protocol An open standard for data-driven blocks

Even if that is true, I could see this being worthwhile. It becomes more of a UI protocol than a framework and there could interop between components built with various frameworks...maybe something like blockprotocol.org

Also in that video though, @ryansolid.bsky.social said that you have to set very explicit expectations on how they communicate for it to work. I'm curious if you would agree with that and if it aligns with what you're describing as "framework-like behavior without the framework"?

React is killing the web YouTube video by Theo - t3․gg

I was watching this youtu.be/LrUpYblQIV4?... and in the chat it was noted that they are composable but there weren't any examples. Though I think @jakelazaroff.com below linked one!

Interesting, has anyone written about this anywhere? I'm curious about how that ends up working in practice. Are all the web components written with the same framework but just put together as web components? If they were cross-framework would that cause problems with state management, perf, etc.?

I’ve been very curious about what happens if you tried to build an application by stitching together web components instead of staying within React/Solid/Svelte/etc., but haven’t had a chance play with it yet. Is this something you’ve tried at all? Would love to hear more if so.

Car driving down the street, on the back it says “BLUESKY GLASS WORKS”. The license plate is “GLAS MAN”.

I see the Bluesky team finally found a way to bring in some revenue.

That plus adding all of the current security concerns with these technologies on top of the city’s horrendous record of data security…recipe for disaster.

I’m bullish on there being good uses of these AI tools to improve services in the long run, but the idea of using AI to improve the efficiency of the city’s terrible service is hilarious. They should try improving the service before focusing on efficiency.

The only way I can see this happening is if someone comes up with a more convenient slang. I would say people don’t call it “the mountain” for correctness, it’s because “the escarpment” is a mouthful. Though at this point I fear it may be too late.

If it becomes a consistent problem you could try enabling “only followers can reply” on your posts to curtail the drive by comments