| 22 Nov 2019 |
|  fmaurer joined the room. | 17:00:56 |
 Taylor Buchanan | There is also only ~500 people who have joined the new version vs ~8500 in the old version | 17:01:53 |
 ox | Ha, that is probably it. | 17:04:12 |
 Matthew | yeah, definitely not placebo | 17:05:17 |
 swedneck:permaweb.io | also matrix.org fixed their performance issues, which probably affects things | 17:07:05 |
| Matthew | the amount of timeline isn't the problem | 17:05:37 |
| Matthew | but the number of users & servers is. | 17:05:41 |
 Andy | In reply to@taylor:taylorbuchanan.com
There is also only ~500 people who have joined the new version vs ~8500 in the old version It seems room upgrades are a nice way to purge members, heh | 17:06:01 |
|  Daniel Caixinha joined the room. | 17:08:53 |
 Valodim | *fixed some of their performance issues | 17:10:19 |
 Aaron | ox: btw this room was v1 before, not v4. Room upgrades don’t happen very often since they are so destructive. | 17:10:38 |
 Smar | I wonder how much amount of users (in a room) affects to performance | 17:10:49 |
| Valodim | quite a lot, especially if those users are on many different servers | 17:14:26 |
| Smar | hmm, performance in terms of delays or cpu/memory usage? | 17:16:09 |
| Valodim | well, both. delay is caused by more cpu and memory usage, it's not like the machines wait around more if there are more users | 17:17:08 |
| Smar | well, with basic logic, it shouldn’t take too much of resources, since you can just listen for input and validate and write it down and after it it works exactly as local users | 17:19:15 |
| Smar | I don’t know how Matrix in reality works though | 17:19:21 |
 Foks | In reply to @smar:yuuko.smar.fi
well, with basic logic, it shouldn’t take too much of resources, since you can just listen for input and validate and write it down and after it it works exactly as local users the listening to input and validating it is the hard part | 17:22:30 |
| Smar | it shouldn’t take that much resources tho... and even if it is the intensive stuff, there is ways to optimize stuff | 17:23:56 |
| Foks | well uhh, get to work optimizing stuff then? | 17:24:21 |
| Smar | too lazy | 17:24:30 |
| Valodim | haha, it's not that simple | 17:24:44 |
 vurpo | In reply to@smar:yuuko.smar.fi
well, with basic logic, it shouldn’t take too much of resources, since you can just listen for input and validate and write it down and after it it works exactly as local users The Matrix server, when a client sends an event (like a message) to a room, first of all needs to decide whether or not the user is allowed to do that or not | 17:25:13 |
|  panicbit joined the room. | 17:25:37 |
| vurpo | To be able to do that, it needs to have the most recent state of the room, in which those rules are written | 17:25:41 |
|  samendacott joined the room. | 17:26:11 |
| Smar | but can’t that data just be kept in memory? it shouldn’t take too many bytes, if all that data contains is just user ids and permission bits | 17:26:37 |
|  himcesjf_ joined the room. | 17:26:41 |
|  jamaliskandar joined the room. | 17:27:20 |
| Valodim | no. this stuff is complicated | 17:28:13 |
