Time Nick Message 00:06 rdococ Strange... I'm not getting Lua error messages for *anything* in 5.0.0. Minetest just crashes. 00:15 IhrFussel I just wiped my entire minetest.conf by accident...now I gotta remember all my values D: 00:20 IhrFussel minetest.conf allows comments right? Like ## this setting is for x 00:20 rubenwardy yes 00:20 IhrFussel Can the comment also be its own line? 00:20 rubenwardy yes 00:21 rubenwardy I'm not sure what the rules are on escaping # 00:21 rubenwardy maybe # only works if it's the first non-whitespace character 00:21 rubenwardy ie: can't use it after a setting 00:21 rubenwardy tl;dr: try it 00:22 IhrFussel Will it crash if it's not supported? Or just ignore the line? 00:22 rubenwardy it'll be included in the setting 00:22 rubenwardy ie'll foo = bar # thing will set foo to "bar # thing" if comments aren't supported after a setting 00:23 rubenwardy lol, "ie'll" 00:35 IhrFussel Huh looks like 'server_dedicated' was removed from conf? I remember having that in my wiped conf 00:36 rubenwardy server_dedicated doesn't need to be set, it's done internally by minetest 00:37 IhrFussel Yeah it's not part of conf.example either 00:44 rubenwardy IhrFussel: you may be interested in some code that CTF has, it precompiles out setting calls into constants 00:44 rubenwardy https://github.com/MT-CTF/capturetheflag/blob/master/build.sh 00:50 rdococ Is there an easy way to change the maximum HP of a player in 5.0.0? 00:51 rubenwardy yes 00:51 rubenwardy max_hp 00:51 rubenwardy well, not easy 00:51 rubenwardy you also need to manually adjust the hp bar to show more hearts 00:51 rdococ It won't show them on its own? 00:52 rubenwardy there are long freezes every 18 seconds :( 00:54 rdococ Interesting... it seems to scale the HP down (so that if you have, say, 100 max HP, one heart is 10 HP instead of just 1 or 2). 00:59 rdococ Increasing HP always just seemed to me like increasing defense, but in a slightly more (or less) complicated way 01:00 IhrFussel rubenwardy, freezes regarding what? Talking about builtin sqlite auth? 01:02 rubenwardy the map save interval is 18 seconds, and every 18 seconds it freezes for about 0.7s 01:02 rubenwardy map saving is disabled though 01:02 rubenwardy actually, I think it may be less than that 01:02 rubenwardy ah, it happens when someone enters or leaves 01:02 rubenwardy looks like it's sban 01:03 rubenwardy well, fix hopefully coming soon ) 01:06 IhrFussel Yes sban is still heavy in your recent profiler report 01:07 Fixer kek 01:09 IhrFussel Regarding the cache I talked to shivajiva again...and he said the conf setting only tells sauth how many accounts to cache at LOAD time... the actual dynamic account cache during runtime is unlimited 01:09 rdococ That's weird. My character died, but didn't have any HP when they respawned. 01:21 IhrFussel fixed_map_seed is useless if you switch to another mapgen...or not? Cause the map will look different either way 02:19 IhrFussel Well...okay letting sauth run without caching for a while reports higher numbers but still...NOTHING compared to the old sauth version (200-300ms) 02:19 IhrFussel - on_prejoinplayer[1] ............................... | 2 | 11799 | 2022 03:15 rdococ I'm getting a problem with register_on_hpchange - I have a function that is guaranteed to either return hp_change or 0, but apparently it's returning nil? 03:15 rdococ well, not nill but a non-number 03:16 rdococ can't seem to reproduce it though 03:52 rdococ How do I determine if two stacks can be fully combined? 03:55 rdococ nvm found 05:11 rdococ Sneak peek... https://i.imgur.com/6US1a0J.png 07:43 sofar https://github.com/minetest-mods/minetest-mods.github.io/issues/72 07:43 sofar open feedback request to anyone 09:09 nerzhul hello, weblate has been updated to master 09:09 nerzhul https://hosted.weblate.org/projects/minetest/minetest/ 09:09 nerzhul let's go 09:23 nerzhul merging #7892 09:23 ShadowBot https://github.com/minetest/minetest/issues/7892 -- Deprecate modpack.txt and use modpack.conf instead by rubenwardy 09:24 nerzhul merging #8013 09:24 ShadowBot https://github.com/minetest/minetest/issues/8013 -- Fix various bugs (Anticheat, Lua helpers) by SmallJoker 09:32 Krock hi imaginary tenplus1 09:32 nerzhul Krock, what the hell ? 09:33 Krock I'm prepared every day but he's now only rarely here 09:33 nerzhul ^^ 09:33 Krock quiet hard to do a race this way 09:35 nerzhul latest android apk in production is now installed on 933 devices over 7k 09:35 nerzhul currently the stats shows -100% crash :p 09:39 nerzhul weblate is now correctly working, nice :=) 09:39 nerzhul many rtanslations levels has lowered to 60% 09:39 Krock didn't they already update recently already? 09:40 Krock perhaps I just messed up updatepo.sh 09:41 nerzhul in fact weblate has its own remote 09:41 nerzhul and you must rebase against it 09:41 nerzhul but it's difficult 09:42 nerzhul generally what i do is: fetch weblate remote, checkout all po files to minetest master, push, run updatepo.sh push 09:42 nerzhul and then i go to weblate interface reset & pull 09:43 nerzhul i know many doesn't like transifex but interface and intergrate with repo are better :p 09:43 nerzhul interetsing comment on the play store: one say he plays minetest on school pc :p 09:50 Krock well, RUN_IN_PLACE builds are portable (Windows) and portable among similar systems (Linux) so it's not surprising :D 09:50 nerzhul maybe it's to study :) 09:50 Krock although an external HDD/SSD is a way better experience than a flash drive 11:12 rdococ Trying to copy a screenshot of Minetest to GIMP causes GIMP to take a long time and then say that there is no image data in the clipboard. 11:15 Krock try a 2nd time? 11:15 Krock I also had that and just tried again to make it work 11:15 rdococ Ah. I'll try it next time I need to get a screenshot. 11:15 rdococ Anyway, a second peek: https://i.imgur.com/tlmsLS0.png 12:25 Krock so after all these years, we again have fireflies as "mobs" in-game 12:33 nerzhul yeah but in lua :p 12:39 rubenwardy Lua isn't slow 12:39 rubenwardy The engine is slow 12:42 rubenwardy with Luajit, the bottlenecks are with communication with the engine 12:42 rubenwardy not with the lua code itself 13:13 celeron55 it's not slow really, it's just often impossible to make it do what would be efficient in a given use case 13:14 celeron55 like a tool box that's missing tools for a job you went to do 13:18 Fixer fireflies are just "mobs", not real mobs 15:54 nerzhul Merged #8060 15:54 ShadowBot https://github.com/minetest/minetest/issues/8060 -- Content store: Disable more details dialog for now by rubenwardy 16:20 nerzhul merging trivial #8061 16:20 ShadowBot https://github.com/minetest/minetest/issues/8061 -- Fix wrong code comment by DS-Minetest 16:21 rubenwardy wrong channel 16:21 rubenwardy :P 16:57 Krock rubenwardy: some work for you: https://forum.minetest.net/memberlist.php?mode=viewprofile&u=24189 16:57 Krock needs ban & delete 16:57 rubenwardy done 16:57 Krock missed one: https://forum.minetest.net/viewtopic.php?f=32&t=21774&p=340192#p340192 16:58 rubenwardy lol what 16:58 Krock ugh.. the post is half-way existing 16:59 rubenwardy weird 16:59 Krock thank you :D 16:59 rubenwardy there's an option to delete the topic luckily 17:14 IhrFussel rubenwardy, is the new sauth version live on CTF yet? 17:18 MinetestSam Yes it is 17:21 IhrFussel And what does /status say? 17:21 IhrFussel I'd assume not constantly 0.5 secs anymore 17:21 IhrFussel Although sban is still a problem 17:22 MinetestSam Haven't checked that but just a second 17:24 MinetestSam IhrFussel, max_lag=0.562125 17:24 IhrFussel Then it must be sban...the lag doesn't come down cause of it yet 17:25 MinetestSam The players are saying that they are having no issues 17:25 IhrFussel But in theory the lags should be less frequent now 17:25 MinetestSam It is afaik 17:25 MinetestSam Less than before 17:26 IhrFussel sban is the only bad mod remaining according to rubenwardy's profiler report he posted last night 17:26 MinetestSam shivajiva was working on it I think 17:27 IhrFussel He is working on it right now...the one he finished was sauth 17:27 IhrFussel And on my server sauth improved from 200-300ms per lookup to 2-3ms (without caching) 17:27 MinetestSam Well good to see he hasn't left yet 17:28 MinetestSam The database also keeps on increasing, so the idea suggested was to delete players with score less than 100 17:28 MinetestSam player database that is 17:29 * Krock wonders what his score is 17:29 * Shara adds a few numbers too it just to mess with things 17:30 Shara to* 17:30 MinetestSam ABMs and Chatplus was also removed so that should have also helped 17:30 MinetestSam Heh 17:31 IhrFussel The amount of records doesn't matter now anymore... shivajiva changed the DB structure to 1. properly INDEX the names column it seems and 2. split the big table with all records into 11 tables like auth_abc auth_def etc 17:32 IhrFussel CTF had 650,000 records IIRC ... now one table will have max 200,000 cause of the splitting 17:33 IhrFussel On my server with 150,000 accounts some tables have 30k records and other have only 5k (auth_misc only has 250 records cause barely anyone chooses a name that starts with - or _) 17:34 MinetestSam That is good to hear 17:35 IhrFussel He said the splitting is not necessary but imrpoves performance further... and I would prefer having multiuple tables with 30-50k records instead of having 1 with 150k 17:35 IhrFussel improves* 17:35 IhrFussel multiple* I should type slower sometimes 17:39 MinetestSam Well I am waiting for ruben to post some stats soon 17:39 IhrFussel Well in his mod profiler report that he ran last night it showed 1µs for sauth 17:40 IhrFussel But I wonder how high it could go when he runs the profiler during busy times 17:40 IhrFussel When CTF is full of people 17:42 Krock the execution time will likely be very similar if it measures the average execution time 17:42 Krock or min/max of each 17:42 MinetestSam At the time of me asking the players about the lag there were 32 players out of 30 so that should give a good estimate 17:44 IhrFussel (this was yesterday with auth caching off) -> - on_prejoinplayer[1] ............................... | 2 | 11799 | 2022 17:44 IhrFussel On my server 17:45 IhrFussel A max of 12ms is absolutely fine 17:46 IhrFussel Assuming builtin sqlite also uses proper indexing...shouldn't the 2 perform very similar now? 18:04 rubenwardy they should 18:14 IhrFussel Question: When does CS_HelloSent appear for the server? on_prejoin() or on_join()? 18:19 Krock rejoin I guess 18:22 IhrFussel Ugh...I thought you FINALLY fixed the fact that new players cannot connect anymore when too many are in the "media phase"... I assumed CS_HelloSent is when the client JOINS the world 18:23 IhrFussel Why do we count players in media phase towards the LIMIT? 18:24 IhrFussel Imagine 10 or more mobile players are stuck there...cool now they block connecting for anyone else 18:25 IhrFussel It looks like you even WANT it like that according to nerzhul's PR here https://github.com/nerzhul/minetest/commit/60171295309d64319a2b39464e5b2a3e8bc7dd06#diff-27330fe8b12f281b2b6d288458912306 18:27 nerzhul it was fixed 18:28 IhrFussel Where? It wasn't backported then 18:31 nerzhul in 0.4.x ? 18:34 IhrFussel The only commit I find is where you changed CS_active to CS_HelloSent 18:36 IhrFussel No wait... you changed it from CS_Created to CS_HelloSent but if I read it correctly both of them happen before the media phase 18:37 IhrFussel So it didn't fix this major issue where players count toward the limit as soon as they pass auth 18:37 nerzhul yep 18:37 nerzhul it's to solve the problem that non connected client can steal server slots 18:38 nerzhul media phase is after yep 18:38 IhrFussel That was one issue...the other is still existent...please move the check if too many users are connected to before join 18:38 IhrFussel I mean only count clients that joined 18:39 nerzhul it's the goal yes 18:40 IhrFussel So if 10 mobile users are stuck for minutes in media download they don't count until they told the server that they are ready...and if the limit was reached until then they are disconnected 18:40 rubenwardy it should probably be a little bit more flexible 18:42 rubenwardy like: give loading players a tentative slot. They lose this slot if they take too long or if a player that lost their slot finishes loading. New players can't start loading unless (slots + tentative_slots) < max_users 18:42 rubenwardy actually, it would be simpler to kick maybe 18:43 IhrFussel Also...isn't it (in theory) still possible to make a bot that is still able to fill slots? Maybe more difficult cause they have to emulate more steps but nonetheless possible 18:43 rubenwardy the issue with just moving the slots to after media loading is that media loading can be a resource drain 18:44 rubenwardy and it's also annoying to spend time loading and then be rejected 18:44 IhrFussel The media cache helps there 18:44 rubenwardy really it would be nice for a server to disconnect clients which don't use remote media 18:44 IhrFussel The next time they load it will not take that long anymore 18:45 rubenwardy ie: so you can guarantee server-based sending isn't happening 18:45 rubenwardy oh, I guess 18:45 rubenwardy well - the proposed check after load is probably better if you have limits on media sending too 18:45 rubenwardy better than current 18:45 rubenwardy heh 18:45 rubenwardy it's all a bit shit really 18:47 IhrFussel On mobile media loading can take 1 or 2 minutes on weaker devices (I know it did take that long on my server in the past) 18:47 IhrFussel I mean without downloading 18:47 IhrFussel I heard it's cause of missing VRAM 18:48 IhrFussel Sorry I mean media & nodes init together 19:03 rubenwardy fun commit: https://github.com/rubenwardy/minetest/commit/8babb15bb90cea2a5717f0e90eaacfb45b652ee6 19:04 rubenwardy unfortunately you can't just respond with "this file doesn't exist" 19:05 rubenwardy would probably be more useful to log this 19:07 Krock auto -> const auto & 19:07 Krock and that'll be a very long list 19:23 Krock o/ spotted an active tenplus1 on the forums 19:25 rubenwardy I did 19:37 Fixer due to the ongoi... sigh, can you stop telling me that? 19:38 IhrFussel rubenwardy, btw does your profiler still run on CTF? Any new values available? 19:38 rubenwardy I disabled it 19:38 rubenwardy What's the profiler overhead? 19:39 IhrFussel I think it barely lags the server more... I cannot confirm a max_lag increase of more than 0.1 19:39 IhrFussel But it#s probably better to wait for the fix I told you in msg 20:46 IhrFussel rubenwardy, do not update to the new dev version of sauth now... it's completely broken and he tries to fix it 20:47 IhrFussel Connecting isn't even possible cause of "2019-01-06 21:39:18: ERROR[Main]: ServerError: AsyncErr: ServerThread::run Lua: Authentication handler didn't return privilege table" 20:50 IhrFussel Looks like it only breaks when you enable the caching 21:17 IhrFussel Looks like his idea (preloading X entries) exceeds the time the engine allows for the auth handler to setup 21:21 rdococ Hehe 21:22 IhrFussel Why does the engine limit the auth handler setup time? 21:22 rubenwardy it shouldn't 21:22 IhrFussel Or is it a general 'mod startup' limit 21:23 rubenwardy oh maybe it's a race condition 21:23 rubenwardy like, due to singleplayer 21:42 IhrFussel rubenwardy, would this make sense? Somebody connects to the server but sauth is still busy with populating the cache so it cannot return a 'privilege table' in time 21:45 rdococ is there a key to hide the HUD, or take a screenshot without the HUD? 21:45 rubenwardy f2 21:45 rubenwardy or f3 21:46 rdococ F2 hides chat, which I think might be enough for my purposes. 21:47 IhrFussel Okay this explanation makes more sense -> auth handler is used as part of the priv handler ... all priv requests to the core call on auth to supply and modify the stored privs 21:47 IhrFussel And builtin calls auth directly on startup 21:58 rdococ https://i.imgur.com/KC3Gaky.jpg 22:48 luk3yx Is there a way for a CSM to know what the CSM restriction level on the server is? 22:50 sfan5 no 23:04 * luk3yx should open an issue for that