mstdn.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A general-purpose Mastodon server with a 500 character limit. All languages are welcome.

Administered by:

Server stats:

17K
active users

Erin Kissane

Even in early release, the Elk (elk.zone) web app for mastodon is so good, and also no waiting list ✨

Tangentially related, I REALLY miss local search for finding posts that I saw and want to refer back to—like the one introducing Elk—but didn’t bookmark. Is there something that can help with this?

(Please do not tell me to get better at bookmarking.)

@kissane Oh god thank you. I have been utterly unable to use the normal mastodon app in any functional way.

Also my eyeballs

@onewil lol I almost tagged you on the post

@kissane My best workaround for this is Ivory, which lets you search the timeline you’re viewing. But that view doesn’t always have the content you’re looking for.

(Other apps may do this as well, but I don’t remember.)

@jsit I need to get on the list then I guess, ty!

@kissane I ran full text search on my Mastodon instance for a couple years. My servers were churning 24/7 to keep up and it never worked well. Eventually, I gave up.

@smeg @kissane interesting, I would not have expected that, but I've never run a text-search service. I'm curious-- how many users? Do you remember what percentage of resources it took vs. all the other processes? Was the load more from the indexing or searching?

I'm curious because I want to add full text searching to a social media app I'm writing. Thanks! :)

@jamesmarshall @kissane

The go-to method is Elasticsearch. The issue for me was indexing - it was crunching on indexing all of the time, yet when I attempted to search, it wouldn't find content <10 days. My masto instance has only a dozen or so users, but I believe it was also trying to index content from the federated feed (and my instance is well-federated <3), so it was probably an issue of quantity for the CPU power I had.

AWS offers Elasticsearch, so it can be done at higher scale

@smeg @kissane interesting, thanks for your insights. I've been looking at FlexSearch (the JS one, not the C# one-- they look like totally different projects). FlexSearch claims to be much faster, but I haven't tried it yet. Unfortunately, right now it seems to be JS only, which works for me but may not work for other projects.

nextapps-de.github.io/flexsear

flexsearchflexsearchNext-Generation full text search library for Browser and Node.js

@smeg @jamesmarshall @kissane IIRC, bird site had the same issues in the early days (soon after the acquisition of Summize. There were days when search would fall behind by days or even weeks, because it couldn’t index as fast as users were creating new tweets. Eventually, the search team temporarily focused indexing on a subset of active users and off new and less active accounts (80/20-ish rule).

@smeg @jamesmarshall @kissane There were times when the reverse was true too. Older tweets would fall out of the index because of capacity constraints. We take all that for granted now. That is, until we want the same immediate indexing that goes forever back here too.

@kissane probably overkill but that rewind.ai app lets you search through everything you’ve seen on your (apple silicon) Mac. Pretty neat, but I haven’t actually used it for real yet. What you described sounds right up the alley of what to sure going for though!

www.rewind.aiRewindRewind is a personalized AI powered by everything you’ve seen, said, or heard.

@kissane Do you know how Elk is in terms of accessibility features? I really hope they are building inclusively from the start.

@felner I don't think I know enough about overall masto accessibility to say, but I noted that they make alt text easy to spot/lick through to, which seems promising.