So as yall can tell, with this huge influx of users lately, lemmy has gotten really, often unbearably slow. This is entirely my fault for not structuring the database caching (lemmy currently uses a SQL technique called materialized views), in a proper way.

With a small amount of users / posts, it seemed fine, but as we scaled up, this is proving not to be a good method to do caching. These past few days (even with all the adoption), I managed to build a way to do that increases speed by maybe 30%, but its still not good enough, so I’m tackling it from a few different angles. All my other work is on hold, because this seems to me to be the most important thing to handle right now.

In other news, !chapotraphouse@lemmy.ml is in the process of making a lemmy instance, and we’ve seen a few others pop up too. Once we deploy federation, we’ll be happy to federate with them, and link them in an official instance list.

Also we’ve got some movement on people building a lemmy mobile app, as well as @IronOxidizer@lemmy.ml 's building a lemmy-lite, a liteweight non-javascript front end, which when done, we’ll be happy to bundle with lemmy directly (Possibly at lemmy.ml/lite) which would be great to have.

My original ideas were to have a web app (like this one), a native android app written in kotlin + jetpack, and a terminal app like rtv / tuir, so its great to see development here.

Let me know anything I can do to improve the site, and especially be sure to add feature requests to the github. Thanks!

Just like dessalines, I was mostly busy this week handling the huge influx of users, first from Hackernews and then from Reddit. At one point the site was brought down completely because of too many open websocket connections (in other words, too many people were on the site at the same time). Luckily this was rather easy to fix with some changes in the nginx config (which I should probably document somewhere).

Another effect was a huge number of new issues and pull requests that had to be answered and reviewed. Of course contributions are very welcome, but they do take time away from other things.

I also setup Grafana to monitor the server. This still needs some improvement to show better stats (wish I had time to dig into that).

We also came under a DDoS attack. Our previous provider wasnt helpful in dealing with this at all, so we moved to OVH instead. Their DDoS attack detected an attack within half an hour of moving the server, and successfully mitigated it. Since then things have finally calmed down.

Today I finally managed to get some actual coding done (for the first time this week). I made some progress migrating to the new version of our ActivityPub library, but its not that straightforward and still needs more work.

There might be other things I’m forgetting, maybe next week I should keep a list or something.

I love the commmunication you admins have, it’s really nice to know exactly whats going on. You guys rock!

Oops, another one: The subscribe button responds too slowly upon pressing it. Most buttons change to loading circles (e.g. post) or respond immediately (e.g. voting) presumably without waiting for the backend. The subscribe button calls out to the backend (slow) but doesn’t update itself until the backend replies leading to me clicking it multiple times and getting errors.

Wanted to say I love the rss feed for notifications :)

Am curious, do you think it’s necessary to have a big red “banned” label next to username when seeing a list of posts for a community? It’s a little disruptive in cases like this

I really wish titles for external links went to the external link and not the comment section, breaks my expectation of what is going to happen and then I need to click again to view the content

I’m really curious if anyone has thought about trying to use noria as a db for lemmy? It’s a research project - so not really appropriate for real production use - but in theory I think it might significantly improve the performance.

Below the title of posts (e.g. this one), there is a series of buttons. At the right hand end of that series of buttons is a “…”. Clicking that … only expands to one more button.

When the expansion menu only contains one button, maybe replace the “…” with that button.

Edit: That’s all I can think of for now - great work y’all!

deleted by creator

The alert UI was somewhat unintuitive to me. It wasn’t clear how to mark messages as read/why it didn’t go away after I read them. I’m not sure how to fix or even if this was just me and not a common experience.