Time  Nick        Message
07:52 nerzhul     hello, #9226 is ready to be merged
07:53 ShadowBot   https://github.com/minetest/minetest/issues/9226 -- Dockerfile: use alpine to reduce size by tsao-chi
07:53 nerzhul     i amended the MR with my changes from my PR sfan5, and added pg support, that should be fine now
07:59 sfan5       well I don't know much about docker but I guess I can take a look
07:59 nerzhul     if you have docker installed locally: sudo docker build . then take the container id at the end of the build, sudo docker run <id>
07:59 nerzhul     s/container id/image id/
08:01 nerzhul     https://pastebin.com/kbqdNYYx
08:01 nerzhul     here is my postbuild output
08:01 nerzhul     we don't use env vars yet on minetest to configure if i'm correct, no ?
08:02 sfan5       I don't think there are any
08:03 nerzhul     i think about supporting that for world.mt, it will be more container friendly than now, in my usecase i'm blocked by our world.mt file because it's a config file in the data folder which needs to be wrote, and if i configure it it's a read only file
08:04 nerzhul     and it may be systemd more friendly too
08:04 nerzhul     (if you use systemd template units)
08:04 sfan5       I think several distros already package minetest with template units
08:10 nerzhul     fine , then if we add some env vars for the most inportant config keys it can be very nice for end users
08:36 nerzhul     sfan5 fine for #9226 ?
08:36 ShadowBot   https://github.com/minetest/minetest/issues/9226 -- Dockerfile: use alpine to reduce size by tsao-chi
08:43 sfan5       nerzhul: wouldn't the resulting image be identical if you used "COPY . /usr/src/minetest"?
08:43 nerzhul     yep except it will fail on mtg clone if you have in your local workspace a mtg already installed
08:43 nerzhul     and it will copy all your build junk (which is 2GB on mine) to the docker image on each build which is time consuming :D
08:44 nerzhul     the image is built on each commit on gitlab, if we have a problem we will know it very fast if it's your doubt
08:45 sfan5       hm
08:45 sfan5       fine by me then
08:45 nerzhul     and maybe we can also have the image build on github if we use actions if we decide it too :D
08:46 nerzhul     cool
08:47 nerzhul     i'm working on a WorldSettings object which can handle envvars and our world settings code inside the ServerMap constructor (which is not very nice to maintain)
08:49 sfan5       which settings do you plan to put in there?
08:50 nerzhul     the db type, and params, it permits to run a docker container without publishing a volume with write perms and set the config from the host, it's more plug & play
08:50 nerzhul     (read from env vars, superseeding the file)
08:50 nerzhul     and put the config read code outside of servermap, directly inside the WorldSettings object
08:53 nerzhul     oh we now have support for a base map with a writeable map, permitting to reset world on each restart without loosing the whole world ? very nice
08:54 sfan5       the container needs write perms on the world folder anyway so why is world.mt an issue?
08:55 nerzhul     in some env it will be read only, but yes we permit config write failures, reading the current code
08:56 nerzhul     i will propose the PR and we will see, else i will keep it for my tree :)
08:59 nerzhul     sfan5, can i merge the docker PR then ?
09:00 sfan5       yes
09:00 nerzhul     okay merged, thanks
09:43 nerzhul     this worldsettings approach is quite nice, it permits to protect the underlying Settings object and have proper config getters, i like it :D
09:48 Krock       will merge #9423 in a few minutes
09:48 ShadowBot   https://github.com/minetest/minetest/issues/9423 -- GUIFormSpecMenu: Add basic element highlighing debug feature by SmallJoker
09:48 nerzhul     very cool feature !
09:49 Krock       thanks
10:01 Krock       merging
10:21 rubenwardy  Daily PPA reports build failure
10:24 sfan5       link?
10:26 rubenwardy  https://launchpad.net/~minetestdevs/+archive/ubuntu/daily-builds/+recipebuild/2540806/+files/buildlog.txt.gz
10:27 rubenwardy  I can check it out in about an hour, mainly posted here so.i remember
10:27 sfan5       sounds like git on their side somehow broke
10:56 nerzhul     yep sounds to be on their side
11:17 nerzhul     here is my PR to permit to overwrite world.mt configurations from environment: #9647
11:17 ShadowBot   https://github.com/minetest/minetest/issues/9647 -- Add WorldSettings object & env var support by nerzhul
13:01 rubenwardy  is Travis still down?
13:18 sfan5       https://travis-ci.org/github/minetest/minetest/jobs/674011319 looks fine to me
13:46 Krock       rubenwardy: Wuzzy responded. can the PR now be merged?
13:47 Krock       #8873 is basically 4 core devs vs a modder. Shall I wait anyway?
13:47 ShadowBot   https://github.com/minetest/minetest/issues/8873 -- Add default stack size setting by SmallJoker
13:59 sfan5       by tagging it with "user feedback needed" I mostly meant Wuzzy's feedback
13:59 sfan5       but unless someone suggests an alternative approach I don't see how this PR would not get merged
13:59 nerzhul     i just added some missing env vars to #9647 now it's complete
13:59 ShadowBot   https://github.com/minetest/minetest/issues/9647 -- Add WorldSettings object & env var support by nerzhul
13:59 nerzhul     i'm just disappointed to see i didn't backported my postgresql auth backend before leaving :
13:59 nerzhul     i'm just disappointed to see i didn't backported my postgresql auth backend before leaving :D
13:59 nerzhul     it's the most useful part for multiple server sharing same user base
14:01 nephele     I am really confused how this pr got that much discussion .-. , i agree scope granularity would be nice to have, but it really is unrelated to that setting
14:37 DS-minetest Krock: regarding https://github.com/minetest/minetest/pull/9101#discussion_r407176737 (you suggested to remove the FieldSpec entry for the clip element of the scroll container), what should I use as id for the clipper element? maybe -1 or 0?
14:37 Krock       same as for statictext. ID 0 is used for such elements
14:37 Krock       -1 is probably the formspec itself
14:38 DS-minetest ok
14:38 Krock       I did a quick test without those FieldSpec to ensure there's really no side-effects
14:38 Krock       but what counts more is the code, which doesn't do anything if there's no data within the FieldSpec
14:38 Krock       or there's no FieldSpec
14:40 rubenwardy  I'm modifying the read_object_properties to log a deprecation warning for defining an initial object property not in initial_object_properties - this has been deprecated for years, and has caused some confusion
14:40 Krock       and in terms of legacy sorting they'll behave the same as StaticText from labels or field titles
14:40 DS-minetest done
14:40 rubenwardy  the simpliest correct way to do it is to do this: https://rwdy.uk/UbUfv.png
14:41 rubenwardy  and then check the return value here: https://rwdy.uk/RLCWy.png
14:41 rubenwardy  I don't like this
14:41 Krock       that's the only way
14:42 Krock       L193 and L204 can be technically wrong sometimes
14:42 Krock       actually all of them could be wrong in case the supplied value is still the same as the old one
14:42 Krock       good enough IMO
14:43 rubenwardy  well, I'll change it to "was_specified"
14:43 rubenwardy  the alternative is to refactor the read_object_properties to take a   std::map<std::string, PropertySpec>   where PRopertySpec defines a type like color or something like that, but this won't work if there are properties that are interdependent or complex in some way, and makes putting the results from that map to an ObjectProperties more complex
14:44 rubenwardy  I guess    map<string,type> -> read_properties -> map<string, union> -> objectproperties
14:44 rubenwardy  hmm
14:45 Krock       or rather "specified". shorter, and also because it's kinda still specified in the scope where it's used
14:45 rubenwardy  `a_property_was_specified_in_this_table`
14:46 Krock       or that ^
14:47 rubenwardy  the benefit of making it use a map for types is that you can then log a notice with the property name
14:47 DS-minetest can't you move the check for whether there is something (which causes a warning) into a second function that just checks for existence? or is that too much code duplication?
14:47 rubenwardy  but I think the complexity outways the cause
14:48 rubenwardy  DS-minetest: that's another option, that would duplicate names though
14:48 rubenwardy  actually, it's already duplicated
14:49 rubenwardy  if enums could have string values and be iterated over, you could do:     ObjectPropertyLuaNames.HP_MAX    to force new properties to be added to the table, and then add a separate check function
14:57 DS-minetest (there's an EnumString class for such things)
15:00 DS-minetest but you could also make a static const char *const ObjectPropertyLuaNames[NUM_ENUMS] = {...}; and cast the enum values to integer values
15:09 rubenwardy  lol, read_object_properties is broken
15:09 rubenwardy  It takes an index argument, but it only works if the index is equivalent to -1
15:09 rubenwardy  it just happens that in all invocations that's the case
15:29 kaedr       Hey All, just found minetest about a week ago, been playing and enjoying it a lot, wanted to get involved in development.
15:29 kaedr       Been getting set up with visual studio per the instructions on github, ran into an error trying to build the first time: https://pastebin.com/ZebUCaq8
15:30 sfan5       huh well that'll have been my fault
15:31 sfan5       ..or maybe not
15:32 nerzhul     very strange command, you cannot use vscode instead of VS ?
15:32 kaedr       I tried using vscode, but vcpkg wouldn't recognize it
15:33 sfan5       no idea about the error but you can disable the .rc file compiling by removing this line https://github.com/minetest/minetest/blob/master/src/CMakeLists.txt#L457
15:34 sfan5       that should make it work, you lose the window icon though
15:44 rubenwardy  #9650
15:44 ShadowBot   https://github.com/minetest/minetest/issues/9650 -- Log deprecation initial properties by rubenwardy
16:01 rubenwardy  What. Deprecation notices aren't enabled by default?
16:02 rubenwardy  modders are never going to see them
16:02 DS-minetest rubenwardy: wouldn't an array of `const chat *`s suit better for object_property_keys than a vector of std::strings (const values => no need for smart pointers)
16:02 nephele     I always put my handling of deprectated calls to error... but i must say, i only stumbled upon that setting by accident :P
16:02 rubenwardy  raw arrays suck
16:03 rubenwardy  can I iterate through a raw array?
16:04 rubenwardy  like, c++11 style
16:05 DS-minetest yes
16:05 DS-minetest ranged for loops are possible for arrays (just tested)
16:06 rubenwardy  doesn't compile for me
16:06 rubenwardy  because it's an incomplete type
16:07 rubenwardy  moving it to the header works
16:11 rubenwardy  DS-minetest: updated
16:12 DS-minetest (writing the size into the [] would also make it a complete type)
16:17 Krock       you never allocate arrays in headers
16:17 Krock       do that in source files, which are only parsed once
16:18 Krock       same goes for static class members
16:26 DS-minetest are the object properties read every time a new entity is added? if so, wouldn't it make more sense to read an ObjectProperties object per entity name once, in register_entity?
16:27 rubenwardy  Register entity is a Lua function that simply adds the definition to the table
16:28 DS-minetest I see
16:32 nerzhul     rubenwardy: std::array permit that
16:32 rubenwardy  I don't think that std::array allows implicit lengths
16:33 rubenwardy  you'll get a compile error, so could just increase it each time
16:33 DS-minetest length should be 30
16:36 rubenwardy  ok, I'm happy with that
16:44 sfan5       if you don't want to manually increase the length each time you can also use a nullptr-terminated C-style awway
16:45 sfan5       array*
18:52 nerzhul     seems minetest is not very kubernetes friendly
18:52 nerzhul     i only have IPVS load balancing but it seems some packets are lost and i'm blocked on the black login, i can talk ingame but i'm nowhere in the map
18:54 nerzhul     debugging that can make me crazy i think :D
18:55 nerzhul     or maybe it's a ARM issue ?
18:55 nerzhul     calling directly the service on the host seems to do the same
18:56 nerzhul     it reminds me when i tried to rework the network layer
19:15 sfan5       I tested minetestserver on aarch64 just this week and it worked fine
19:19 nerzhul     with a client on x86/amd64 ?
19:20 nerzhul     the server is launched and some packets are working properly but not the init part of the client, the unreliable part it seems
19:21 sfan5       yes
19:23 nerzhul     hmm
19:39 nerzhul     it's strange the packets are received but something may be wrong
19:42 nerzhul     maybe due to the nat or something like this
19:43 nerzhul     only reliable packets seems to work, not non reliable
21:19 Zughy       I thought the menu was made all in Lua, background aside, but even if I disable the whole builtin part, a panel still remains => https://i.imgur.com/eBpRZR9.png . Could someone please point at me where exactly it's generated?
21:23 rubenwardy  That's a formspec window. You can set the color to transparent to disable that
21:23 rubenwardy  bgcolor[#00000000]
21:23 ShadowBot   rubenwardy: Error: Delimiter not found in "HTTP Error 404: Not Found"
21:23 rubenwardy  Notice the 8 digits  not 6
22:27 Zughy       v2u32 is a custom type or..? I can't find anything on the internet but on your repo and I don't see a declaration anywhere
22:28 sfan5       it's a vector of two u32's
22:28 sfan5       the declaration is here tho https://github.com/minetest/minetest/blob/master/src/irr_v2d.h
22:28 sfan5       core::vector is Irrlicht's vector type
22:29 sfan5       s/type/class/
22:29 Zughy       makes sense, ty
23:08 kaedr       So I just finished a build, and it errors out because it's missing zlib1.dll  zlib1d.dll is there, any ideas where I might have gone astray?
23:09 sfan5       check where vcpkg put it?
23:11 kaedr       Ok, cool, copying it in from vcpkg got things working. But shouldn't the build have put it there automatically?
23:13 sfan5       I heard vcpkg + cmake can do that automatically so probably yes(?)
23:19 kaedr       Maybe I'll poke around the cmake config a bit, see if I can figure out how to fix the magic.