Time  Nick        Message
10:42 MinetestBot Krock: Mar-16 04:10 UTC <p_gimeno> Your ascii biome is awesome! but the Z loop is inverted, the map is upside down. https://gist.github.com/SmallJoker/03c92442c4b81a6d2e7573c7950a10cc#file-init_22_ascii_biome-lua-L25
10:42 MinetestBot Krock: Mar-16 04:34 UTC <p_gimeno> it would be nice if it showed the area around the player, instead of around 0,0,0, and is triggered with a chat command
10:42 Krock       o/
10:43 Krock       sfan5: MinetestBot is so wonderful that I got the same message three times
10:51 Emerald2    lmao
11:26 IhrFussel   btw how did you make it so that lava hurts you even when just standing in a 'slab'? I thought damage_per_second only affects you when your head is in it or was that changed in 0.4.17.1?
11:28 IhrFussel   Well it doesn't affect water so either the engine has something hard coded for lava or a mod causes this on my server
11:28 IhrFussel   Let me test this in vanilla MTG
11:29 IhrFussel   Seems to be hard coded
11:30 IhrFussel   Or MTG somehow checks this
11:44 IhrFussel   Ok it's hard coded and if water had a damage_per_second attached to it it would likely also be fired when the feet only are inside
12:02 sfan5       Krock: I may have forgot to implement locking for that part
12:31 IhrFussel   I want my server stats to be more transparent in the game so I changed my /lag to include the amount of lag spikes for the entire server session ... that should tell them whether or not the gameplay is fluent/stable enough
12:32 IhrFussel   A high lag spike is considered '> 0.3 s' and a critical spike is '> 0.5 s'
12:33 IhrFussel   Various WE actions will actually make it look worse than it really is...but no way to prevent that so I'll just accept it
13:53 Krock       #8385
13:53 ShadowBot   https://github.com/minetest/minetest/issues/8385 -- Add /help formspec for commands and privileges by SmallJoker
14:44 ANAND       Krock: Nice! :)
14:54 IhrFussel   Why is there a PR that attempts to limit the debug.txt file size? debug.txt is ESSENTIAL on servers
14:55 Krock       and android players cry about the size
14:56 IhrFussel   Server owners need debug.txt to check recent PMs and dig/place/inventory actions ... that is required to investigate rule breaking and other unwanted activities
14:56 IhrFussel   Without debug.txt and with rollback disabled servers have no way anymore to check who griefed where
14:56 rubenwardy  debug.txt should be limited by default
14:57 rubenwardy  server owners should then be able to disable the limit, and use proper rotation
14:57 IhrFussel   The limit should not be 1 MB by default
14:57 IhrFussel   1 MB is a joke
14:57 rubenwardy  for most users, that's fine
14:58 rubenwardy  because they'd only want to give the last 20 lines of it when it crashes
14:58 rubenwardy  it's probably a good idea for the limit to depend on whether the server is in dedicated mode
14:58 IhrFussel   So you expect server owners to magically know that they need to enable/disable a flag in their settings to actually log useful stuff for more than a few hours?
14:59 rubenwardy  what do you suggest we do?
14:59 rubenwardy  because users are currently running out of storage, especially on mobile devices
15:00 IhrFussel   Disable limit by default or make it at least large enough so that server logs don't miss too recent stuff in case the owner forgot to change their settings
15:00 IhrFussel   Or tell server owners somehow ingame/in menu that the log is limited currently and how to change that
15:00 rubenwardy  I don't support disabling by default
15:00 rubenwardy  but I would support enabling by default on Android, and disabled elsewhere
15:02 IhrFussel   Message in menu won't work cause many servers run without GUI
15:04 IhrFussel   rubenwardy, then Win/Linux users could still complain ... I think either set the limit to something bigger like 50 MB or so (even the cheapest Android devices have a storage of 4 GB these days) or just find a way to tell the server owner (and only them) that the log is limited to X MB currently
15:04 IhrFussel   Possibly when they join their server
15:04 ANAND       The PR could make use of different logfile max sizes depending on whether or not the `ENABLE_SERVER` cmake flag is set. Would that provide the much-needed compromise between 1MB logs for players and a little more unrestricted logfile size for servers?
15:07 IhrFussel   ANAND, there are some server owners that don't need a huge debug.txt, but on many 'serious' ones that have rules and use every method possible to enforce them (griefing, stealing, forbidden protections, illegal contents in PMs, brute-force attacks and many more) the log is one of the most important things when running a server
15:09 IhrFussel   I actually already limit mine to the most recent 3 million lines
15:09 IhrFussel   Which is about 250 MB
15:10 IhrFussel   And those 3 million lines still only go back to ~ 10 days
15:10 IhrFussel   It depends on how popular a server is
15:11 rubenwardy  50MB is massive for Android
15:11 rubenwardy  could be a variable size
15:12 IhrFussel   That means if someone reports something that happened more than 10 days ago all I can say is 'sorry dude, too late now to check'
15:13 IhrFussel   What I would rather like (but that would also be a lot more complex I imagine) is not a size or lines limit...but a date limit
15:13 IhrFussel   Like 'only log everything to 14 days ago'
15:15 IhrFussel   Cause the size/lines don't tell you how many days back the log will go ... on a server that#s less popular the same amount of lines could go back to 1 month while on a very popular server the lines could only go back to a few days
15:15 IhrFussel   Then the default could be 1 or 2 days
15:24 IhrFussel   Also many servers have mods enabled that flood the log by default... like hoppers/pipeworks and similar mods that fake players and do actions as the fake player
15:27 red-045     Pretty sure you could make it easily date based if you logged to DB
15:27 ANAND       Good point, IhrFussel
15:28 red-045     IhrFussel: have you considered setting up a DB for logging important events?
15:30 IhrFussel   Wouldn't a DB require about the same size?
15:31 IhrFussel   I mean 3 million lines in a DB cannot be much smaller than 3 million lines in a .txt file I assume
15:31 IhrFussel   The important events to log are the ones that appear the most
15:32 IhrFussel   node place/dig, inventory actions, PMs
15:32 IhrFussel   I think by default even the CHAT is logged in debug.txt
15:33 IhrFussel   I already moved that one out of debug.txt and to its own file
15:34 IhrFussel   On a busy server it's quite common to have 10-15 lines per SECOND
15:35 IhrFussel   I forgot...there is 1 type of messages that appear a lot but are not needed (at least not in my case) and those are when players punch entities
15:36 IhrFussel   Those lines
15:36 IhrFussel   2019-03-16 16:36:11: ACTION[Server]: valo56 right-clicks object -14899: LuaEntitySAO at (-3540,1.7,-3433)
15:36 IhrFussel   2019-03-16 16:36:12: ACTION[Server]: ololosh punches object -14885: LuaEntitySAO at (1056.64,-3.03831,1034.24)
15:37 IhrFussel   I'm not sure if those are actually helpful
15:37 IhrFussel   The mobs on my server have up to 2,500 HP and one hit by players can cause up to 50 HP damage ... now imagine how many lines appear just for 1 player to kill 1 mob
15:38 IhrFussel   Not to mention that I as server owner cannot really know what 'object -14899' is
15:39 IhrFussel   Let me check how many lines of the current debug.txt are such entity punch lines
15:39 Krock       you could convert that to an u32 and read it back from core.luaentities
15:40 Krock       then you can see what it is
15:40 IhrFussel   Almost 500,000 lines out of the 3 million are ^
15:41 IhrFussel   What I mean is are these lines really important? Could they maybe made optional?
15:42 Krock       log level to INFO would be posible
15:43 IhrFussel   But would it distinguish between entity and player? I mean player -> player damage should be in ACTION
15:44 IhrFussel   Cause that is helpful
15:46 IhrFussel   I guess someone could propose to move those lines to INFO and if denied at least add an opt-out flag
15:47 IhrFussel   These lines only really matter on servers with mobs...with lots of mobs
15:50 Krock       yes, that's possible
15:50 Krock       (player damaging player -> actionstream, otherwise infostream)
15:51 Krock       btw: it is possible to punch yourself
15:51 IhrFussel   Yeah but AFAIK that will also log as you punched you
15:51 Krock       just make it so that the client aborts reading its own entity definition and a dummy object will appear
15:52 Krock       yes. player X punched player X
15:52 Krock       I'd push a commit to do that, but I'd like to have #8350 merged before that
15:52 ShadowBot   https://github.com/minetest/minetest/issues/8350 -- Damage: Play no damage sound when immortal by SmallJoker
15:54 IhrFussel   No rush...but thanks for attempting it =) It will only affect 5.1.0 either way I'm guessing
16:01 Krock       exactly
16:16 benrob0329  Are there any alternative circuitry mods?
16:17 rubenwardy  elepower
16:17 rubenwardy  !mod elepower
16:17 MinetestBot rubenwardy: Elepower - A new powerful modpack [elepower] by IcyDiamond - https://forum.minetest.net/viewtopic.php?t=20320
16:17 rubenwardy  supports node_io, which is really nice
16:17 rubenwardy  node_io is a standard for pipeworks/mesecons/technic/etc like mods
16:17 rubenwardy  it means you can add machines without depending on the mod which implements the circuitary
16:18 rubenwardy  like, in capialism_game, I use it to implement fractional distillation
16:20 benrob0329  rubenwardy: a circuitry mod, not just for automation. Judging by the fact that it advertises mesecons support for "toggling your automation" that tells me its probably not a circuitry mod
16:21 rubenwardy  hmmm
16:21 rubenwardy  I was pretty sure that elepower had its own circuits
16:21 rubenwardy  but supported mesecons too
16:21 benrob0329  I'll look into it
17:05 benrob0329  rubenwardy: yeah this is more like s technic replacement than anything else.
17:31 red-045     If you used a DB it would be easier to keep exactly 10 days of logs
18:20 tenplus1    Hey folks :)
18:20 tenplus1    Hi Krock
18:21 tenplus1    hi MIner
18:25 VanessaE    good morning twelveminus1
18:25 VanessaE    :)
18:25 tenplus1    ehehe... hey Vanessa :)
18:25 tenplus1    tomorrow is the big switchover day
18:25 VanessaE    it is?
18:25 * tenplus1  is nervous
18:26 tenplus1    Xanadu is moving over to Minetest 5.0 ... all mods tweaked and tested through the week
18:26 VanessaE    Oh G*d oh shit oh no PANIC PANIC PANIC *runs around in circles*
18:26 VanessaE    ;)
18:26 tenplus1    lol
18:27 Fixer       i will panic if LazyJ moves LG2 to 5.0
18:27 sofar       I wouldn't on a sunday
18:27 sofar       do it during the week - monday
18:27 tenplus1    hi Fixer :P hey Sofar
18:27 Fixer       hi
18:27 sofar       then if problems arise, you have a few days until the weekend
18:27 tenplus1    I cannot, Shinji is off Sunday and works through the week
18:27 Fixer       was there a 5.0.1 release?
18:27 sofar       nope
18:27 Fixer       tenplus1: i would rather wait for 5.0.1
18:28 VanessaE    no need to wait :)
18:28 sofar       there's no reason, really
18:28 tenplus1    I've been running a standalone version of Xanadu using 5.0 and all mods for a week... so far all is working a-ok, no crashes
18:28 VanessaE    5.0.0 works well enough (as long as you use unified inventory to avoid the inventory bug)
18:28 tenplus1    inventory bug ?
18:28 rubenwardy  you only need to wait if    (LOTS_OF_ITEMS && !UNIFIED_INVENTORY)
18:29 rubenwardy  typically people who like bloat also use unified inventory
18:29 tenplus1    hi ruben
18:29 VanessaE    lots of items != bloat.  it's called "choices". :)
18:29 tenplus1    we're still using Inventory Plus :) mwhehehe
18:30 rubenwardy  well actually, only if in creative too
18:31 tenplus1    what's this inventory bug ?
18:32 rubenwardy  it crashes if the string form of a detached inventory is over 2^16 bytes = 65356
18:33 rubenwardy  which happens in creative mode with the MTG creative mode when there's a lot of items
18:33 tenplus1    ahhh, that's not a big deal then
18:33 tenplus1    is limited circumstances
18:35 tenplus1    the new Tundra biome is quite weird btw...
18:54 benrob0329  Hello tenplus1
18:54 tenplus1    hi benrob o/
18:55 rubenwardy  tenplus1: new feature I'm working on https://user-images.githubusercontent.com/2122943/54468060-280ffb00-4781-11e9-810b-763ee623bd55.png
18:55 rubenwardy  #8383
18:55 ShadowBot   https://github.com/minetest/minetest/issues/8383 -- Add formspec styling using a style[] tag by rubenwardy
18:55 Krock       o/ tenplus1
18:55 tenplus1    yaay, that looks a lot better :) great feature
18:56 Krock       late greet but still
18:56 tenplus1    always wanter coloured formspec buttons :D
18:56 tenplus1    ehehe
18:56 Krock       rubenwardy: it's very nice (seen the PR) but I wonder how much we want to try to make the formspecs better
18:56 rubenwardy  that's true
18:57 rubenwardy  fact is though: no formspec replacement has appeared yet
18:57 rubenwardy  most people want something that exists
18:57 tenplus1    they do the job well enough, would be nice to make them more customizable
18:57 tenplus1    colours are a good start for certain elements :D
18:57 tenplus1    2d drawing would be nice also
18:57 Krock       we'd also need plenty of work to implement the special buttons and lists
18:58 tenplus1    or the ability to show entities inside formspec
18:58 rubenwardy  image_button and item_image_buttons are just buttons
18:58 Krock       or a html viewer in a formspec
18:58 Krock       or running a vm which runs electron to show a webpage using a mod in minetest
18:58 rubenwardy  so I could hypothetically reuse the work I already have for them
18:58 rubenwardy  heh
18:58 tenplus1    heh
19:00 tenplus1    yeh saw that earlier :)
19:01 tenplus1    mesecons is an amazing mod but oh so slow on servers
19:01 benrob0329  It really depends on what people do with it
19:01 tenplus1    yeah... must be a way to speed things up though
19:02 benrob0329  If you have a really fast clock that's constantly updating then yes it'll be super slow
19:02 Krock       *le click*
19:02 benrob0329  But if you're not being stupid with your circuits it runs fine (for me anyways)
19:03 rubenwardy  Krock: I'm not planning on it being fully functional, and probably won't implement all the option
19:04 rubenwardy  but it's something that allows these to implemented without breakages: #5385 and #5361
19:04 ShadowBot   https://github.com/minetest/minetest/issues/5385 -- FormSpec: support custom colors in dropdown by adelcoding1
19:04 ShadowBot   https://github.com/minetest/minetest/issues/5361 -- FormSpec : Add background color for textarea/field/pwdfield by adelcoding1
19:07 rubenwardy  I think that it's better to view it as a framework to allow styling elements without breaking them on older clients, and then to judge each customisation one by one
19:07 rubenwardy  I find coloring buttons very useful for marking dangerous actions
19:07 tenplus1    same as when listring[] was added
19:07 tenplus1    no breakage
19:08 rubenwardy  the issue is, however, that in its current form it may cause problems when/if Irrlicht makes a new releas
19:09 tenplus1    would a new version change how colours are selected and drawn ?
19:09 rubenwardy  no, it could result in compile errors if they change how the elements work
19:09 tenplus1    damn
19:10 benrob0329  That sounds like an Irrlicht problem tbh
19:10 rubenwardy  it is
19:10 rubenwardy  but it becomes our problem
19:10 rubenwardy  Irrlicht is a perfect example of how *not* to do C++
19:10 tenplus1    arent we already using the latest irrlicht ?
19:10 rubenwardy  well, depends on what they have installed on their system
19:11 tenplus1    hmph
19:12 tenplus1    this makes it all the harder to keep things running and add new stuff
19:12 rubenwardy  in Irrlicht, you can't just extend an element
19:12 rubenwardy  you have to copy the whole thing
19:12 rubenwardy  it sucks so bad
19:13 tenplus1    forking a feature to improve a feature... would end up massive
19:13 rubenwardy  yeah
19:13 rubenwardy  it's ridiculous
19:13 red-045     you can't override a method or what's the issue?
19:14 rubenwardy  red-045: things are hidden behind an interface
19:14 tenplus1    does formspec allow for graphical buttons to use [colorize ?
19:14 red-045     ah
19:14 rubenwardy  so you can't just subclass the button as you have no header for it
19:14 rubenwardy  tenplus1: yes
19:14 Krock       tenplus1: text can already be colorized using minetest.conlorize
19:14 Krock       -typos
19:14 Krock       inb4 there's also support for background color, but it's broken
19:14 tenplus1    not so much text but buttons using a bitmap background that can be coloured also
19:15 Krock       ah
19:15 rubenwardy  there's already support for background color?
19:15 tenplus1    it's a stand-in feature
19:15 rubenwardy  pretty sure not
19:15 Krock       rubenwardy: well, the code exists
19:15 rubenwardy  the code doesn't exist, because I had to implement it
19:15 Krock       https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L2351
19:15 Krock       speaking of text background color
19:15 rubenwardy  "The escape sequence sets the background of the whole text element to
19:15 rubenwardy  `color`. Only defined for item descriptions and tooltips"
19:16 Krock       inb4 I cannot read
19:16 tenplus1    image_button[0,0;3,1;"bg.png^[colorize:ff00ff00";mybutton;violetbutton]
19:16 Krock       sorry, I though that was already a thing but broken
19:17 Krock       tenplus1: try using formspec_escape on the middle part
19:17 tenplus1    ah
19:17 Krock       surely not quotation marks
19:17 tenplus1    was quick example :P
19:18 Krock       I yet only used it for image[] but it should work elsewhere as well
19:24 tenplus1    "entity_image[0.5,0.5,3,5,character.b3d,texture]"
19:26 tenplus1    hi fussel
19:26 IhrFussel   https://github.com/minetest/minetest/pull/8382#issuecomment-473571559
19:27 IhrFussel   Almost every server owner uses debug.txt to check which player did or sent what
19:27 IhrFussel   At least the ones I know of
19:27 IhrFussel   Hi tenplus1
19:27 tenplus1    last comment is a good idea...
19:27 tenplus1    limit size but compress older one's... that way it's archived for use
19:28 Calinou     would be nice if you could plug Minetest into a dedicated logging platform, it could work better for large servers
19:28 Calinou     https://github.com/grafana/loki
19:28 tenplus1    hi Cal
19:29 Calinou     I have yet to play with it (or Prometheus/Grafana for that matter)
19:29 IhrFussel   If rollback was more efficient maybe more servers would switch to that instead...but debug.txt is the only easy solution when you need to check where someone griefed/stole etc
19:33 tenplus1    logrotate is a good way to go, keeps both parties happy
19:33 VanessaE    rollback's main problem is that the engine lacks any way to prevent some actions from being logged (i.e. technic automation), so the log can grow massive, quickly (tens of gigs in a month)
19:34 tenplus1    yup, that's why on every restart I archive log and reset, keeps things simple
19:34 tenplus1    hi cwz
19:34 rubenwardy  https://i.rubenwardy.com/3AtI6.jpg
19:35 tenplus1    lolol
19:35 VanessaE    I used to trim the log to just the last month's worth of entries via a couple of sqlite commands, but that stopped being adequate when the *trimmed* log started staying in the 20 GB range
19:36 IhrFussel   Maybe MT should log the chat and PMs to a separate file ... cause those are not really 'actions'
19:36 tenplus1    I was looking for a way or command that records the last 100 lines of a terminal window and stores in a .txt file instread of logging at all
19:38 red-045     would be great if someone made a proper audit system for servers
19:38 rubenwardy  defintiely
19:38 tenplus1    wait, doesnt Cat5Tv have such a thing for their own server ?
19:38 red-045     do they?
19:38 benrob0329  A script that runs through debug.txt and properly logs it would be nice
19:38 tenplus1    lemmie find the name of it... they have a monitoring system all setup they made
19:39 tenplus1    think it was called NEMS or something
19:40 benrob0329  NEMS Linux is a general server monitoring OS
19:40 benrob0329  Iirc
19:40 tenplus1    they did something with it...
19:40 tenplus1    lol, gotta ask robbie, cant remember now
19:41 * tenplus1  stress eats
19:44 IhrFussel   Well Krock said earlier he will move the 'player was punched by object [somenumber]' to INFO which should decrease log size a lot already on busy server with lots of mobs
19:45 tenplus1    yeah, every dig, punch etc. adds up
19:46 IhrFussel   Those need to stay...but I think nobody needs to know (unless developing) that a player got punched by an entity or that a player punched an entity ... those spam the log sometimes more than dig/place
19:47 IhrFussel   Spawn 5 or more hostile mobs around a player...now every sec 5 lines appear in debug.txt
19:48 Krock       IhrFussel: only with the log level set to action or higher
19:48 IhrFussel   Now imagine a server with 20-30 players and several of them fight against mobs in this very moment
19:49 IhrFussel   Krock, not sure I understand ... right now those messages appear with default log level which is ACTION right?
19:49 Krock       yes. if they were changed to INFO, then they're hidden
19:49 Krock       i.e. no longer logged
19:49 tenplus1    agreed
19:49 sofar       that's why I prefer logging to systemd - it automatically sizes my log files to a reasonable size so I don't have to worry about rotating or overgrown log files
19:49 IhrFussel   That's what I mean
19:50 IhrFussel   Move those messages to INFO so that they are hidden from the default log level
19:50 tenplus1    how'd you do that sofar ?
19:50 sofar       just have MT log to stdout when you run it from a server, and not to debug.txt
19:51 tenplus1    ooh interesting
19:51 tenplus1    thanks
19:51 sofar       iow don't set --logfile
19:51 sofar       journalctl -u minetest.service -> read minetest's log files
19:52 sofar       another reason I prefer it that way is so that I can see the relation between other system events better
19:53 sofar       e.g. mtmediasrv shows someone connecting, a little later minetest shows the login
19:53 tenplus1    you could add a forum thread on how to do this for server setup... would be handy :) and informative
20:09 tenplus1    nite folks o/
20:09 Fixer       \o
20:24 Calinou     yeah, journalctl is more powerful than plain text logging
20:31 * Fixer     corrupts Calinou's journal
21:03 Sokomine    rollback is great. wish that'd be more easily available. could even help in local building...
21:20 Peppy       hi
21:21 Peppy       being able to restrict rollback to a selected zone would be very nice too