https://atproto.com/guides/identity#example-self-hosted

Example: Self-hosted#

Let's consider a self-hosting scenario. If using did:plc, it would look something like:

• The handle: alice.com
• The DID: did:plc:12345
• The hosting service: https://alice.com

However, if the self-hoster is confident they will retain ownership of the domain name, they can use did:web instead of did:plc:

• The handle: alice.com
• The DID: did:web:alice.com
  The hosting service: https://alice.com

So if I'm understanding this right, the DID is supposed to be the permanent identifier of the person and always forever be associated with the content they post in bsky.

If I have this right, then does this mean that if you use did:web instead of did:plc that your id will always permanently be tied to the domain name you choose, and that you can never change your domain name and still keep your content?

But furthermore, from my reading here, did:plc is supposed to be "temporary" (thus the name), so assuming some suitable DID method is added to the standard, how does this affect users? How do they migrate to this new standard method if their DID is supposed to be static forever?

Is there any plan for the future for atproto to include a standardised ability for clients to host local plugins in a wasm/webassembly sandbox?

well... not yet. you'll get something like this if you send an unknown lexicon to

Hopefully soon though.

In reply to

moved to @shreyan:beeper.com (@shreyanjain:matrix.org)
totally! domain usernames aren't actually linked to the pds, so there's nothing stopping you 🙂

I was actually referring to

It appears there is actually no restriction whatever on labels, save for length:

https://stackoverflow.com/questions/25199202/which-rfcs-specify-the-syntax-of-hostnames-and-constraints-on-dns-hostnames

Hm, yeah, at least with regards to mastodon, you're right, mastodon.social has the most users, but it's not super concentrated. I wonder how many of that user count is active.

But I think

This probably has a lot to do with discoverability and defaults (e.g. I think Element just defaulted to

In reply to

1@10allday:matrix.org
How can we ensure that you have to pay to access any server on the Bluesky network? What norms will need to be accepted to ensure that this happens and models after Usenet subscriptions?

It's an interesting question how little you can get away with labeling.

At first glance anything by anyone who one of your users followed would need labels or worse the full content of your Indexer but actually it may be much less than that. Most content is boring and no one will ever see it rise to the level of their timeline or search results. We may be able to only run the labelers over stuff that has enough base rank that it is likely to be recommended

In reply to

bnewbold
hi folks, popping in quickly to say: we miss you! has been a wild week and have not had time to keep up here. we have even paused all github interactions for a week. we intended to land a round of dev docs and info about federation around now, but things are getting pushed back b/c we have urgent moderation needs. keep an eye out for that soon. we can't build this whole ecosystem ourselves, and know that we are not doing well w/r/t docs and support right now, but hope to get our attention back to it soon

In reply to

@jonaharagon.com
why would we want it to do that?

In reply to

moved to @shreyan:beeper.com (@shreyanjain:matrix.org)
thats the beauty of atproto, things dont need to be built into the protocol. anybody can just write new lexicons?

In reply to

moved to @shreyan:beeper.com (@shreyanjain:matrix.org)
so, exactly what they do right now?

In reply to

1@10allday:matrix.org
Yes, but with browser-like features that JavaScript uses to communicate with the page.

They do, ActivityPub has a similar concepts to lexicons called ~~ActivityStreams~~ Activity Vocabularies. Despite this ActivityPub has multiple clients such as Mastodon, Misskey, Peertube, exc. that work with eachother fine

https://www.w3.org/TR/activitystreams-core/
https://www.w3.org/TR/activitystreams-vocabulary/

Just a tip for the future, this is not how human beings interact with each other. If you want people to engage with you, you should try to avoid rudely demanding their time and attention. If no one responds to your question, take the hint and don't push it.

"PAY ATTENTION TO ME AND VALIDATE MY OPINIONS" is behavior that even my 4 year old has learned doesn't work

In reply to

d@dead10ck:dead10ck.com

Just a tip for the future, this is not how human beings interact with each other. If you want people to engage with you, you should try to avoid rudely demanding their time and attention. If no one responds to your question, take the hint and don't push it.

"PAY ATTENTION TO ME AND VALIDATE MY OPINIONS" is behavior that even my 4 year old has learned doesn't work

On a dark day when you find yourself 4 ssh tunnels into the production server your trying to debug and the Java Flight Recorder (JFR) will only output the metric you need by running the metric server webUI.

Can you manage to plum port 8088 all the way back to your laptop?

Will that open a major security hole?

Do you even know how to get a command to the SSH sessions in the middle of your terrible SSH chaining?

And then you remember Lynx can read the web and show you text. The ancient ones have sent you a gift from the past and have saved your present.

In reply to

moved to @shreyan:beeper.com (@shreyanjain:matrix.org)
for sure, although in a pinch curl will do

sends a POST request to https://yourpds/xrpc/com.atproto.repo.createRecord with json that looks something like this:

123456789{
        "collection" => "app.bsky.graph.follow",
        "repo" => session.did,
        "record" => {
          "subject" => resolve_handle(session.pds, username)["did"],
          "createdAt" => DateTime.now.iso8601(3),
          "$type" => "app.bsky.graph.follow"
        }
      }

In reply to

mikuhl
I am pretty sure twitter prioritizes the replies that are the OP of the post you are looking at, and then probably by likes or something.

Ah interesting, thanks! I'm not really concerned about BSky's policy, I always figure fear mongering like that is at least part exaggeration, part out of context, etc, and regardless, being a decentralized system, if you don't like their policy, you can use a different instance if you really want to

But I am curious what happens legally to content that ends up on their indexer from other federated instances, how much of their tos applies, if any, since any user on a different indexer couldn't have reasonably been expected to agree to such terms

It may be better to answer the why is it cheaper as a more general
AtProto vs https
or
content centric vs connection centric

In a protocol with connection centric security you authenticate who you are connected to and then ask questions. If I read

Vs

A content centric security you authenticate the content. The commits to the repo are signed. When I try to load a repo I could pull it from anyone that has it.
For someone like Twitter they no longer need to pay for operating PoPs or an expensive CDN. They can use whatever cheapest co-lo server they can get in a region as a content cache MUCH cheaper. There is no trust in the caches just blobs.

Bonus Points: Twitter is not the only one that could catch the data. If any ISP wanted to give a good users experience they could add their own caching. Users could add their own caching. Client could use P2P tech to add community caching.

The whole content distribution chain has radically different economics.

Troll joined
Well… that sounds ominous. 😂

In reply to

yousefed
Hi! Is there support for permissions / authorization in at proto yet? E.g. allow certain users to read / write specific objects / stores? What are the best resources on this?

Sorry, I just joined a couple days ago. I could read your messages if you have a permalink.

But also, not really sure what user names have to do with gradually increasing the user base in a controlled manner?