Time Nick Message 14:14 grorp luatic: Can you please finish your review of #13964? 🙏 14:14 ShadowBot https://github.com/minetest/minetest/issues/13964 -- Add Lua API function to resolve node/collision/selection boxes by grorp 14:14 grorp It's been three months. 14:29 MTDiscord grorp: ouch, sorry, lost track of it (this was a review I started before I was core dev, so I didn't mark it anywhere..) 14:38 grorp luatic: I see... 14:38 grorp The actual problem is not this specific case, the actual problem is that it always takes multiple months to get non-bugfix PRs merged. 14:38 grorp well, not always, but usually 14:39 MTDiscord The problem is an overwhelming amount of PRs currently 14:39 MTDiscord The only way I can see us tackle that is if we convert more competent, trusted active contributors to core devs 14:40 grorp however, review/merge delays weren't shorter when there were like 80 PRs 14:42 grorp I agree with more core devs as a solution. but the quality of merged code shouldn't suffer either :\ 14:43 MTDiscord 80 vs 100 isn't much of a difference. It's about the trend: Can we handle PRs faster than they pop up? And currently the answer is no. 14:43 MTDiscord And if that doesn't change, delays will necessarily go up. 14:45 grorp a solution could be that core devs are exclusively reviewers, so that they don't contribute to the PR heap and do more reviewing (obviously wouldn't work) 15:19 [MTMatrix] I mean, if it helps we got rid of dual wielding + 2 related PRs since there was no activity... so a big thing, for better or for worse, gone 15:35 celeron55_ it's not a good idea to avoid making PRs as a core dev - core devs are one of the people who will have the best ideas and solutions for features 15:39 grorp this was no serious suggestion 15:40 rubenwardy I always feel guilty opening PRs and try to review some first, probably irrational 15:42 grorp I can relate to that 17:43 LandarVargan How common is it to get "Ran out of sequence numbers!" in the verbose logs? 17:45 celeron55_ you need to purchase the sequence numbers DLC 17:50 Krock or try to get a hacked Minetest version from Fitgirl 17:51 Krock technical point of view: it just means that a packet queue is full. probably because the peer does not receive them in time 17:52 Krock this likely means package loss, which is not nice but not much can be done about that 17:52 Krock s/package/packet/ 17:53 celeron55_ this species of network animal is genetically very close to the disconnect animal 17:54 Krock bench the player. he's not useful. 17:55 LandarVargan So the sequence numbers are per-peer then? 17:56 Krock no, per-channel. there's currently 2 channels per peer. 17:57 Krock which logically means that channel 1 and channel 2 packets aren't guaranteed to arrive in sequence 17:58 LandarVargan Ah, ok 17:58 Krock in practice, the server might process player movement packets before punching, even if the player clicked before moved. 18:14 sfan5 in short: if the player is experiencing packet loss or a disconnect, that's it 18:15 sfan5 or a mod might be hammering the engine too hard 18:15 sfan5 or a yet undiscovered engine bug 18:15 sfan5 (note as of 5.9.0-dev connection with severe interruptions should disconnect quite quickly) 18:15 sfan5 (please test any bugs you might want to report on that first) 18:19 LandarVargan I suspect someone discovered another DoS exploit, but there's a chance I've just managed to break something without realizing. Still looking around through the logs 18:31 MTDiscord If "sequence numbers" is the thing we use for sound or particle IDs, the way they were allocated was changed recently to fix a bug. 18:32 ROllerozxa I assume sequence numbers refers to the networking code 18:33 MTDiscord Hmm, I thought there was an internal sequence generator for handle IDs, which was changed in https://github.com/minetest/minetest/pull/14059 18:34 MTDiscord the idea being instead of reusing the lowest free one, it should keep counting and then wrap around at the end ... but who knows how long since wrapping around was actually tested? 18:34 MTDiscord if it's specific to networking ... well, I guess that's just the pain of reinventing SCTP then I guess. 18:37 sfan5 it is the latter 19:22 LandarVargan Is walls of `Server::peerAdded(): peer->id=` (Presumably for every online player) every second normal? 19:25 LandarVargan Alongside `Server: Handling peer change: id=, timeout=0` followed by `Server: Handling peer change: id=, timeout=1` in the same pattern I think 19:27 LandarVargan The server eventually just stops accepting connections (and the irc mod's relay times out), and I have to kill the minetestserver process to restart it 19:31 sfan5 no, you should update 19:33 grorp merging #13964 in 10 min. thanks for reviewing, luatic. 19:33 ShadowBot https://github.com/minetest/minetest/issues/13964 -- Add Lua API function to resolve node/collision/selection boxes by grorp 19:35 MTDiscord np, LGTM. I'd merge #14330 in 15 min then now that the corresponding Irrlicht changes have been merged. 19:35 ShadowBot https://github.com/minetest/minetest/issues/14330 -- Support absent scene node names by appgurueu 19:35 sfan5 it will not work 19:36 MTDiscord sfan5: I suppose I would need to create an IrrlichtMT release? 19:37 MTDiscord (for the CI to work) 19:38 MTDiscord I mean we could also wait for gltf and just have everyone check out irrlichtmt v14 till then 19:38 sfan5 I'd prefer that 19:38 sfan5 and yes you'd need to create a release on the irrlichtmt repo and upload the right artifacts with the right names 19:41 MTDiscord alright 19:42 sfan5 fwiw you could already merged them now if you put and #if around it 19:45 sfan5 LandarVargan: (assuming you're not running a version state from about Jan 20) 19:55 proller why irrlicht still not added as submodule ? 19:57 MTDiscord proller: it is planned to merge it into the minetest tree soon (TM) 19:57 ROllerozxa just merge irrlichtmt already 20:07 nrz yep just merge and clearnup will be done overtime 20:15 LandarVargan I'm running latest dev with a few added commits like the object observers PR, I'm currently waiting to see if I have issues with raw minetest-dev