Avatar
someone
9fec72d579baaa772af9e71e638b529215721ace6e0f8320725ecbf9f77f85b1

I wrote a little bit about them here

nostr:note1p4yn9mazy3acfpuu7a3ncf4tls3prk8ef68qk23t056xe3scvc0qqmypsd

nostr.mom does that. I don't add it to nos.lol because I want it to be more inclusive.

Replying to Avatar someone

There are a lot of people losing contacts when they try new apps. This lady is one example:

nostr:note1g65yepvkgtwry47nuwqpzyt35mk5w5yjw4je2k4pljr4cv7qp0hsemptfc

The reason may be:

Apps have to have default relays for them to start communicating with the network. This is just there to facilitate things to jump start.

When people use an app for the first time, the app doesn't know user's relay preferences and follow list. So when the app starts with its default relay list and immediately posts those to relays as kind 3, the app overwrites user's preferences. Since the app didn't read user's follow list first, it thinks the user doesn't have any follows. All the follow list is overwritten and the other apps will see that empty follows. If it deletes the previous kind=3 note, it is deleted from the relays as well. The user has to rebuild it from scratch or from a backup.

Another theory may be an app is using kind=10002 for relay list but user's relays were actually part of kind=3. Miscommunication between apps may result in such cases.

As a temporary solution I think:

- All the apps should write to 10002 and 3 for a while for relay prefs. 3 for being compatible for previous design.

- 3 and 10003 for follow list.

- If a person follows more than 1000 people they can use 30003 and store several pages of contacts. Maybe save the first 1000 in 10003 and all of them in 30003. If another app is not supporting 30003 then at least they will see first 1000.

- 30003 may be more efficient too in terms of traffic. If they follow like a 1000 people but divide the list by 10, then whenever they follow or unfollow the app doesn't have to do a huge list update but just update a smaller list (1/10th of the whole thing).

kind 30003 tag: "d", "page01" {1/10 th of contacts in this note}

kind 30003 tag: "d", "page02" {1/10 th of contacts in this note}

....

kind 30003 tag: "d", "page10" {1/10 th of contacts in this note}

Another theory is default relays of an app may not include relay prefs of a user. The intersection of those two sets of relays is empty. Hence the app does not read user's prefs at all. This is harder to solve. Relays that specialize in just hosting 1, 3, 10002, 10003, 30003 like purplepag.es can be a neat solution.. Or having more relays...

Enough touching servers! Time to touch grass

Trying https://hosthatch.com/

Gives more ram compared to Hetzner.

There are a lot of people losing contacts when they try new apps. This lady is one example:

nostr:note1g65yepvkgtwry47nuwqpzyt35mk5w5yjw4je2k4pljr4cv7qp0hsemptfc

The reason may be:

Apps have to have default relays for them to start communicating with the network. This is just there to facilitate things to jump start.

When people use an app for the first time, the app doesn't know user's relay preferences and follow list. So when the app starts with its default relay list and immediately posts those to relays as kind 3, the app overwrites user's preferences. Since the app didn't read user's follow list first, it thinks the user doesn't have any follows. All the follow list is overwritten and the other apps will see that empty follows. If it deletes the previous kind=3 note, it is deleted from the relays as well. The user has to rebuild it from scratch or from a backup.

Another theory may be an app is using kind=10002 for relay list but user's relays were actually part of kind=3. Miscommunication between apps may result in such cases.

As a temporary solution I think:

- All the apps should write to 10002 and 3 for a while for relay prefs. 3 for being compatible for previous design.

- 3 and 10003 for follow list.

- If a person follows more than 1000 people they can use 30003 and store several pages of contacts. Maybe save the first 1000 in 10003 and all of them in 30003. If another app is not supporting 30003 then at least they will see first 1000.

- 30003 may be more efficient too in terms of traffic. If they follow like a 1000 people but divide the list by 10, then whenever they follow or unfollow the app doesn't have to do a huge list update but just update a smaller list (1/10th of the whole thing).

kind 30003 tag: "d", "page01" {1/10 th of contacts in this note}

kind 30003 tag: "d", "page02" {1/10 th of contacts in this note}

....

kind 30003 tag: "d", "page10" {1/10 th of contacts in this note}

Another theory is default relays of an app may not include relay prefs of a user. The intersection of those two sets of relays is empty. Hence the app does not read user's prefs at all. This is harder to solve. Relays that specialize in just hosting 1, 3, 10002, 10003, 30003 like purplepag.es can be a neat solution.. Or having more relays...

do you remember if you tried a new client/app/website?

What do y'all think about free to use relays but pay for features (like higher limits, ability to backup, download, pushing to other relays, .. . )?

Right now every note are fetched from multiple relays and this causes a lot of traffic and hot phones. :)

With negentropy a client can only fetch a few notes from some relay, and other few notes from other relay, only the ones that it doesn't already have..

Decentralization is fine but there has to be a copying hub that connects all those. Otherwise people will lose notes. (Not everyone will see every note) nostr:npub1yxprsscnjw2e6myxz73mmzvnqw5kvzd5ffjya9ecjypc5l0gvgksh8qud4 is working on a "hub" thing.

It does speed up after sending to background. Maybe that follower count is taking up too much space?

Amethyst is super cool but especially slow when you go into a user's profile. It becomes almost unusable after seeing a profile. nostr:npub1gcxzte5zlkncx26j68ez60fzkvtkm9e0vrwdcvsjakxf9mu9qewqlfnj5z

Replying to Avatar nym

Are you using AI only or some editing too?

This is a really big deal!!

Vitor broke the whole web 😄

As far as I understand the javascript at the link loads an HTML page, a CSS file and another javascript file all hosted on Nostr relays and renders a web page.

nostr:note16xg8qf6tvjagry2rjy28af7egyvwwl5590kedzuw4zp2epcte25qjl425r

Sorry for making you mad. I trust you and don't have to google. But yeah we have to start thinking about if anything is real or how to check it.

what do you think about this: kind 30053 meaning nostr name system (NNS).

tags:

[['d', 'nos.lol.nns.'], ['A', '88.198.51.48'], ['AAAA','ip:v6']]

in case there are multiple servers available in different continents:

[['d', 'nostr.mom.nns.'],

['A', '5.78.89.174'],

['A', '88.198.51.252']]

Then client can do pings and connect whatever it wants.

Yet another interesting project could be AI trained on nostr wisdom to counteract the undiscerning nature of current models.

clients should not save their default relay list (kind 3) on relays at first login. clients should not save contacts of user (also, kind 3) at first login because that list is empty on day 1. the kind 3 has both of those on the same note and might be causing problems because of this I think. instead clients on day 1 need to make sure they are fetching the follow list from relays. and that should be the last saved follow list from relays saved by previous clients. then if they want they can add the relays to that same note and post back to relays. another remedy could be using 10002 as follow list and 3 for relay list but i am not sure if that is popular.

Relay traffic is 2.5x more compared to June average. Nostr is winning.

how does lmdb directly use the disk?

They may. I use IPs for the above mentioned reasons. The nature of Nostr makes it unable to judge spammers based on pubkeys only. Because account creation is free, I need to look at something costlier to block them, like an IP address.

I haven't changed spam filter codes on the relays for a while. So I can talk about what they are doing:

😸 nos.lol: Looks at pubkeys while determining whether to accept or reject notes, unless the pubkeys from single IP becomes too many. Then it blocks the IP. Does not care about the reports of other people. Suitable for relay - relay syncs as well as regular users, however if the relay is spammy its IP will be rejected. More allowing than nostr.mom, a lot more people are using it, which means a lot more notes. One of the most popular relays!

🤱 nostr.mom: Looks at IPs to determine if someone is spamming. This is more brutal than looking at pubkeys. Hence, less spam. Also looks at reports of other people, so if someone is reported a lot then the notes from that person is rejected. However if the person is well known it would require a lot more reports for him to be rejected. Suitable for regular users who want a cleaner Nostr experience. Not many clients are using it as default relay, so if someone is here that means they may have wanted to be here.

I told my mom about this project. She said go for it as soon as I started. Not sure if she understood it though. Maybe she got an intuition?

Nostr may outlive Apple