In @gossip, I often have chicken-and-egg problem:
- it cant fetch anything until I manually enter some relays
- relays used are picked based on follows list, if the follows list is empty, it does not connect to a relay
- even after I add relays, it often has difficult getting my meta, or my follows list
The first two points are as designed
* The first because I don't want to hardcode any relays for about a half dozen reasons, including favortism, maintenance as relays go offline forever, tending towards centralization, maintaining some association with their operators, etc.
* The second because, as there is no "global" feed, there is nothing to pull down when you aren't following anybody. So why connect to a relay when you don't need anything from it. Unless of course you have configured 'read' relays in which case will connect to those watching for events tagging you.
The third point is troublesome, but I don't know the cause of the particular problem, whether that data is not at the relay or something else. But if you find that gossip is buggy in some regard, I would certainly like to know so I can fix it.
Thanks for replying!
The relay selection in Gossip is great.
I also understand that there are some deliberate choices to enable working with no ID imported as well; keeping follows and relay lists even with no identity.
Used to other clients, I expect that importing an identity leads to showing the follow list and feed, without manually entering relays. If this is not possible by design, I can live with that (then my trick is to keep multiple gossip data dirs and rotating them).
Currently I cannot say if something is buggy or by design, and frankly I have not spent the time to systematically test it; sometimes it works, sometimes it does not. But e.g. now after ID import and manually adding my 8 relays it found my metadata (sometimes that’s sporadic too), but not my follower list (it was updated recently),
I usually import only pubkey, I’m not sure if that’s relevant.
So you are restarting from scratch as you switch identities? I can see how that would be difficult. There probably are some bootstrapping issues where you need to restart for it to update and work properly that I never fixed because I rarely bootstrap (most people only do it once).
Keeping multiple data diretories is a good idea. You might want to symlink (if your filesystem supports it) the cache folder so you don't download the same data multiple times.
