Oddbean new post about | logout
 This is not occuring when I log in to #Primal. Currently posting from #Amethyst
nostr:nevent1qqsz80a4xr4qju2sv8qsyj0xmz9qnn7m3ucy355ftwrffa5zu9t25hcpr4mhxue69uhkummnw3ezucnfw33k76twv4ezuum0vd5kzmp0qgsfye65t5530wq0wpllmdz23luhjxp9drhhh2syae6kk8cp6n3k3zsrqsqqqqqp240726 
 Yeah them niggas be creeping, stay woke

 
 It may be the number of follows. Anything over 2000 follows creates a very large filter for relays and many are now blocking the request.  
 that sounds quite problematic 
 It has been a problem for quite some time. Outbox model eases it, but in the end, free relays can only do so much...  
 I would be happy to pay monthly or yearly to get better functionality.

So far my experience with paid vs free relays hasn't made a noticable difference.

I find the selection of which relays to use and exlcude a confusing process(though I appreciate the descriptions now included in amethyst). 

When adjusting relays I generally just troubleshoot by fiddling randomly until something changes and then leave it alone. I would like to better understand how to optimize relays.  
 yeah, there is so much to do to make that better. :( 

Hopefully we can solve it without having to put our own centralized caching server online...  
 We are early. Thanks for your work man.

It was good meeting you in Nashville 🤙 
 Whatever you did it's working now. Thanks man.

I'd zap you if I had a lightning wallet 😅. All my zaps are forwarded to you though so I hope you're stacking them 🙂 
 See if today's v0.89.9 works. I think we got a bit better. 
 Obtanium hasn't caught it yet. I imagine it'll update soon.

What would you recommend as the best way to update the app? 
 Hum.. mine is already there. Strange, it should be all set. 🤔 
 Mine seems to lag for some reason.

Using this link to fetch updates

https://f-droid.org/packages/com.vitorpamplona.amethyst 
 ohh, no, add the link to the Github directly. FDroid takes 2 days to 2 weeks to update their catalog.
https://blob.satellite.earth/2bd7e308c1797d64fca09b1d61e9bde24c68dd45e501c7383eff1e85392df11f 
 Nice man thanks for taking the time to send the video!!! 
 This is how you install amethyst without the play store. 
nostr:nevent1qqsfxxh5zcmw38n7xrxwmc57yn4f3yh4sk4jg08fhm0rjxd4cvc656cpr4mhxue69uhkummnw3ezucnfw33k76twv4ezuum0vd5kzmp0qgsyvrp9u6p0mfur9dfdru3d853tx9mdjuhkphxuxgfwmryja7zsvhqrqsqqqqqp4t2yfh 
 i don't offer a free relay with 0 spam and no asian languages

poster.place  
 Confirmed. Let me see what I can do...

Relay onNotice wss://nostr.wine/, ERROR: bad req: total filter items too large
Relay onNotice wss://relay.damus.io/, ERROR: bad req: total filter items too large
Relay onNotice wss://relay.damus.io/, ERROR: bad req: total filter items too large
Relay onNotice wss://nostr.wine/, ERROR: bad req: total filter items too large
Relay onNotice wss://nostr.bitcoiner.social/, ERROR: bad req: total filter items too large
Relay onNotice wss://nostr.mom/, ERROR: bad req: total filter items too large
Relay onNotice wss://nostr.oxtr.dev/, ERROR: bad req: total filter items too large
Relay onNotice wss://nos.lol/, ERROR: bad req: total filter items too large 
 I would love to be able to maintain a higher follow count if that can be fixed 🙏 
 Looks like this filter size limit is currently hardcoded in Strfry. 

if (buf.size() > 65535) throw herr("total filter items too large");

https://github.com/hoytech/strfry/blob/32a367738c6db7430780058c4a6c98b271af73b2/src/filters.h#L38 
 I wonder if it isn't better to send the notice, but just crop the amount of filters to the max limit of the relay. 

But I don't know.. I feel like the filter array already allows the relay to run one after another instead of all at once. In that case, it might not make sense to put limits in size. An OR in 2000 pubkeys seems better than 2 subs with 1000 pubkeys each. 
 Yeah not sure! I’ll ask Doug what he thinks is best for performance. 
 it's likely because you follow 2000+ people. unless the client does outbox with a diverse set of relays, or query batching, your timeline query is likely failing. 
 damus "fixed" this with query batching, but its an ugly hack.