Nostr View

Silberengel @Silberengel - 4d

Relays are fast, but a little birdy has informed me, that they're soon going to be even-faster. It will effectively be real-time communication. When warp-speed relays?

1
0
3

Silberengel @Silberengel - 4d

I think most users don't even know how fast relays can be because of the relays they're using. They keep claiming about performance, and it's like,

1
0
3

Silberengel @Silberengel - 4d

It's the clients that are slow, not the relays. If you just write to-and-from relays directly, from the command line, it's basically instantaneous. But you refresh on some clients, and it's like thinking, thinking, thinking...

1
0
3

Silberengel @Silberengel - 4d

Think about it. I'm on the opposite side of the globe, from nostr:npub10npj3gydmv40m70ehemmal6vsdyfl7tewgvz043g54p0x23y0s8qzztl5h and nostr:npub1qdjn8j4gwgmkj3k5un775nq6q3q7mguv5tvajstmkdsqdja2havq03fqm7 and their servers, but it still only takes 9 seconds or less, for me to download Jane Eyre, in its entirety (nearly 200k words and 41 events), render it as HTML, and display it with section navigation. None of that is cached or paginated. That is straight loading of the whole damned thing, and I live out in the boonies with DSL and cheap Internet. That's 0.2 seconds per event download, and the client isn't even running on C++, yes. Could probably half that, with some tweaking. Most people take longer to press the "Enter" button, than it takes to post a note.

1
0
3

Silberengel @Silberengel - 4d

Anyone on the West Coast, with some fiberglass, who can tell us how long it takes, there? If it's a similar time, that means almost all the time is in the client. Otherwise, it would have to be noticeably faster.

0
0
3

Silberengel @Silberengel - 4d

SQL is better for detailed queries, obviously, but most people filter the same things over and over. I.e. "give me the next ten items from my favorite relays". There's been a move to make the filtering more advanced, but this seems like overloaded the scope of the relays with a wannabe RDB. Pull it into an RDB, in parallel, so that it can be queried, as-needed, but don't try to RDB-style everything.

0
0
3

Silberengel @Silberengel - 4d

As you say, they are reducing client-side suck, not relay lag-time.

0
0
3

Silberengel @Silberengel - 4d

I think Ditto does something like that, with Postgres, and Will keeps talking about that sort of thing.

0
0
3

Silberengel @Silberengel - 4d

And personal relays will probably stay small enough, for a while, or even indefinitely. Especially, if it's a mobile relay, like #Citrine, and it eventually calls from the SSD. It's the server-side relays that might eventually need configuration. Especially the ones storing big documents or aggregating, like ours.

0
0
3

Silberengel @Silberengel - 4d

Yeah, the logic isn't efficient. Open an app up and it starts whirring.

0
0
3

Silberengel @Silberengel - 4d

Yeah, that's solid.

0
0
3
Showing page 1 of 1 pages