different clients can have their own design choices.

what about twitter threads? these are very long.

you're wrong. pagination is the hardest problem of the internet. try to get to the page of the first commit of any reasonably large github repository for example.

the future I imagine for relays is that we'll have specialized relays for themes, closed communities, paid or free, with different approaches to antispam etc; -- and also relays dedicated to storing and returning historical data or relays that only store events for some seconds.

my relay implementations (wss://relayer.fiatjaf.com, https://expensive-relay.fiatjaf.com) limits the number of returned posts to the 100 most recent events matching the requested filter. I thought it would be on the interest of each relay to set up some sensible limits.

branle shows only the first n posts on the home page, but stores everything from everybody you follow locally on indexeddb. then paginating back to the past is done using the local database only.

we'll steal that later for a nostr web of trust. I think something like this would work nicely: https://github.com/fiatjaf/nostr/issues/20#issuecomment-913027389

why my relays are having websocket errors?

the process must be improved for more user control over relay interaction, but this is a start: 

testing wss://expensive-relay.fiatjaf.com

testing out wss://btcpay.kukks.org/nostr/ws

testing out wss://nostr-relay.wlvs.space

you have no idea how much I hate translations

I can see it but from wss://nostr-pub.wellorder.net

these people who got banned from twitter and went to places like substack are so bad now. on twitter they were forced to say a thing and say it quickly. now they have room for a lot of boring fillers that we have to click to open, and they put images on them too!

this is great to hear!

ooo what? this is great to hear! in clojure?

nice, I'll look at that. thank you.

I've also seen something done in matrix, cactus I believe, but it didn't look like it fit well the matrix idea.

and even though these comments were made inside branle they are addressed directly to the "main source", i.e. the blog post event itself published by the original author -- instead of what happens today (many people tweet links to a blog post and people comment on these tweets).

what do you mean by "branle is not for blog post"? there is a client-side character limit on every note created on branle, but that doesn't prevent branle from displaying larger notes. and I think it is better for everybody is comments on a blog post can be made from branle.

once on the client, domain.name/post/etc would query multiple (hardcoded) nostr relays to search for replies to the post as kind1 notes. and offer a built-in interface for users to comment on just that note as nostr events.

these would all be viewable from clients like branle.