this post was submitted on 22 Aug 2023
786 points (95.8% liked)
Fediverse
17688 readers
4 users here now
A community dedicated to fediverse news and discussion.
Fediverse is a portmanteau of "federation" and "universe".
Getting started on Fediverse;
- What is the fediverse?
- Fediverse Platforms
- How to run your own community
founded 5 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
That's really cool work! It's a bit beyond my pay grade, so I can't really comment too much about it.
I had a look at the PR you mentioned, and again, while I can't comment on the contents because I am a little out of my depth, may I voice my opinion on the exchange? This is coming from a place of trying to help, since I really do appreciate all the work you've put in and are putting in, and the fediverse can really use your talents, so I hope I don't offend you.
From my reading, it didn't appear that you were being ignored/hazed, and it seemed like the devs would have been open to your improvements. From working and leading big teams, I've noticed that communication and managing emotions is often much harder than writing code. In the thread, it appeared that communication had broken down on both sides (and seemed to have been the case in prior interactions too). Since you mentioned your struggles with autism in the thread, I wonder if that played a part in the tone of the devs perhaps being misinterpreted ? This is, of course only my interpretation, and I could be completely wrong.
Ultimately Lemmy itself is an example of trying to build a community and consensus amongst a broad and diverse group of people, who will often not see eye to eye.
In any case I would like to say I personally appreciate your hard work and really do hope you're able to help make Lemmy better. Thank you!
Can you explain to me why it isn't social hazing?
Do you know how to read a SQL statement? I just can't grasp how it isn't social hazing. I've been reading SQL statements for decades, this is obviously a problematic one.
Can you offer alternate explanations of how 3 people could think that SQL statement isn't ... poor performing and gong to cause problems? And how an SQL statement without a WHERE clause took them months to discover and fix?
Extreme hazing is my best answer. I just can't accept that the SQL statements don't speak for themselves along with the server crashes. 57K users for 1300 servers is very... taking several seconds to load 10 posts....
Look at the date... May... this has been going on since May. If it isn't social hazing ... what is it? I keep asking myself that.
Like I said, this was my interpretation based on reading that exchange. It's difficult to convey tone or intention with text, but I didn't detect hostility from the devs, but I did sense that they were frustrated that process wasn't being followed. Perhaps they should not have gotten hung up on that, but it didn't appear to be malicious.
I do, and your arguments about the joins being problematic seemed solid. From having worked on systems with huge scale, I also agree that Lemmy doesn't seem to be big enough to be brought to its knees by the volume of posts it's processing. However, I'm far from an expert, so I don't want to suggest any certainty about the root causes, especially as I don't have the energy or inclination to dig as deep into it as I would to form that opinion.
I don't know why they weren't receptive, but perhaps they themselves felt attacked. I know that wasn't your intention, but misunderstanding happen, especially over text.
Here, you can dig into what posted days before the pull request you read:
https://github.com/LemmyNet/lemmy/issues/2877#issuecomment-1685314733
June 4:
Given that more than 8 JOIN statements is something PostgreSQL specifically concerns itself with (join_collapse_limit). I hand-edit the query with a single IN clause and the performance problem disappears. 8 full seconds becomes less than 200ms against 5,431,043 posts. And that 200ms is still high, as I was extremely over-reaching with "LIMIT 1000" in case the end-user went wild with blocking lists or some other filtering before reaching the final "LIMIT 10". When I change it to "LIMIT 20" in the subquery, it drops almost in half to 115ms... still meeting the needs of the outer "LIMIT 10" by double. More of the core query filtering can be put into the IN subquery, as we aren't dealing with more than 500 length pages (currently limited to 50).
If it isn't social hazing, then what is going on here? Why has this issue gone on since May and servers are crashing every day?