Time  Nick       Message
00:26 hmmmm      paramat PTAL   https://github.com/kwolekr/minetest/commit/46d46a0535189a9e7468b89bddf101d2b79e409c
05:43 eriix      I've been writing an LMDB backend for Minetest, and I was wondering if anyone would have any interest in this.  It presently works very well for my needs, but I've only tested on AMD64 Linux.
09:18 Calinou    LMDB?
09:23 kahrl      it's the storage backend that openldap uses by default (I think) since recently
09:24 kahrl      previously it was bdb
09:27 celeron55  why are there infinite databases and why does someone want to use each and all of them with a _game_?
09:28 kahrl      databases are well-studied in academia
09:28 kahrl      perhaps writing a database is a popular subject for final-year projects and theses
09:31 kahrl      but yeah, the question about using it for a game still stands :P
09:33 kahrl      what I would like to see is a ldap-based authentication handler for minetest
09:33 kahrl      should be pretty useful for corporate minetest servers
10:30 paramat    hmmmm code looks good
10:47 paramat    please any reviews for #4163 ?
10:47 ShadowBot  https://github.com/minetest/minetest/issues/4163 -- Sky: Darker, bluer sky and improved horizon haze at night by paramat
11:08 paramat    ^ kahrl nore sfan5
11:14 kahrl      paramat: weird, on my monitor it looks like the PR makes the midnight sky lighter, not darker
11:17 paramat    it's a little darker, but certainly bluer. you will need a darkened room to judge those screenshots
11:18 kahrl      true. It's noon here so it's a bit difficult
11:21 celeron55  the dark end gamma of monitors tends to be literally completely all over the place
11:21 celeron55  it would be better to visualize night by making things monochrome or something instead of very dark
11:25 paramat    my night sky here is monochrome, it's maximum saturation blue, and it's fairly light to look like atmosphere lit by full moon
11:35 paramat    but of course the horizon differs. i find the horizon needs a little haze to look right, just as in daytime
11:51 celeron55  the horizon is supposed to have roughly the same color as the clouds in order to fit the clouds on the sky to begin with
11:52 celeron55  when there are no clouds, it doesn't need the hazy color, but generally there are
12:06 paramat    yes. i discovered that day horizon has the same hue as day sky, just less saturation so lighter. so i did the same for night horizon
15:09 paramat    will merge #4185 in a moment
15:09 ShadowBot  https://github.com/minetest/minetest/issues/4185 -- Biomes: Add biome-definable riverbed material and depth by paramat
15:10 hmmmm      aah hold on
15:10 hmmmm      you realize that PRs need to be re-reviewed when a change is suggested and a revision is supposedly made, right?
15:12 paramat    ah ok, docs updated
15:12 hmmmm      the message I wanted to get across with the documentation update isn't quite there
15:13 hmmmm      you're saying specific things about specific fields and giving specific details about future plans in a documentation piece on an interface
15:14 hmmmm      just say that the whole biome API is still unstable and subject to change
15:14 hmmmm      it's also possible for rivers to not become biomes
15:14 hmmmm      who knows?
15:14 paramat    ok. no logs for mapgen channel, that's what i remembered
15:15 hmmmm      don't you have IRC client logs?
15:15 paramat    erm maybe
15:15 hmmmm      alright despite the mapgen channel being decent in terms of filtering out interruptions and keeping the discussion focused, maybe it's not the best because of the whole logging issue
15:17 paramat    indeed. and no i don't have any client history
15:18 hmmmm      https://paste.fedoraproject.org/374861/51398981/
15:19 paramat    ok i remembered wrong
15:19 paramat    thanks
15:20 paramat    +1 for your commit (but untested). it might conflict with 4185
15:21 hmmmm      why don't you commit yours first and then i'll commit mine
15:21 hmmmm      i can probably handle merge conflicts a little bit better
15:23 paramat    ok. my PR is bigger too
15:32 paramat    updated
15:35 paramat    merging
15:35 hmmmm      and IS subject to change
15:35 hmmmm      [11:10 AM] <hmmmm> you realize that PRs need to be re-reviewed when a change is suggested and a revision is supposedly made, right?
15:35 hmmmm      Biome should be capitalized in that sentence
15:35 est31      hmmmm, you are sorta right with that sorta not
15:36 est31      I mean if you ask sb to add a . to the end of some documentation sentencee
15:36 est31      and they update their 1000 line PR
15:36 est31      then you dont have to go through the whole 1000 lines
15:36 est31      nor do you need to wait for an "okay, you placed the dot well"
15:37 est31      it is different when the requirements are more opaque
15:37 hmmmm      no, you're supposed to review the difference between revisions
15:37 est31      like "it needs to be faster"
15:38 hmmmm      how do you know that somebody hasn't slipped in an execl("rm", "-rf", "/");?
15:39 est31      I build on trust
15:39 hmmmm      ideally, all code should reviewed at all times before being merged
15:39 est31      If I see that some people indeed have such behaviour
15:39 hmmmm      and tested
15:39 est31      then I re-review the whole thing again
15:39 est31      hmmmm, the problem is just about time
15:40 est31      also, hmmmm do you do this yourself when reviewing PRS
15:40 hmmmm      yes
15:40 est31      do you download each revision of a PR then check the diff
15:40 est31      and do you check it also when its rebased
15:40 hmmmm      i don't download each revision, no, but i do look at the things that have been changed
15:41 hmmmm      our current setup makes reviewing each exact diff more difficult
15:41 hmmmm      fwiw there's a Gerrit plugin for this
15:41 est31      hmmmm, how do you know which things have been changed if you don't download each revision
15:41 est31      github doesnt support looking at the diff between different pr revisions
15:42 hmmmm      so wouldn't the better question be, "how can we make github support this type of feature"
15:42 hmmmm      or "how could we add this functionality on"
15:44 paramat    the 'is' is earier in the sentence, and i don't think biome should be capitalised
15:44 hmmmm      why not?
15:44 paramat    'is X and Y'
15:45 hmmmm      yeah i saw that when i re-looked at it but what about the capitalization thing
15:45 paramat    well it could be considered a name Biome API, or just a biome API
15:46 paramat    i'm already about to merge so can fix it later, it's not a big error
15:49 hmmmm      when would later be?
15:49 paramat    merged, wow down to 125 PRs
15:50 paramat    after now :]
15:54 paramat    maybe i'll do a cleanup of lua_api, it needs it
15:54 hmmmm      i don't get why you'd agree that something is wrong, and then merge it anyway, and decide to fix it after
15:55 est31      in some times that way of doing things is a good idea
15:55 est31      but it is a very very bad idea if it affects compatibility in any way
15:56 est31      so if you have to fix it you need to break compatibility
15:56 paramat    i'll do a cleanup later today then, as my punishment
15:57 paramat    i considered the capitalisation debatable
15:57 paramat    either is ok, but i'll change it to make it more of a 'name'
15:58 hmmmm      i never thought of it that way, about the name vs. generic thing argument, but capitalized titles for things like that are absolutely standard in technical documentation
15:59 paramat    yeah it's probably a little better
16:00 paramat    also i had gone through the whole merging process again and wasn't going to let a debatable single letter stop me :]
16:01 hmmmm      at where i work to make a commit you need to 'git push review <branch>'
16:01 hmmmm      this runs a hook and it does some kind of special magic that i'm not totally sure on
16:01 hmmmm      but it shows up as a new revision in your equivalent to the PR page
16:01 hmmmm      when a new revision is made, it loses all prior approval
16:01 hmmmm      code review, unit tests passing, etc.
16:02 hmmmm      so every single time you fix one god damn letter or add a period or something, EVERYTHING needs to be done again
16:02 hmmmm      you need to go around to all the same people and beg them to re-add their review +1s
16:02 est31      and this is super annoying to everybody
16:02 hmmmm      and then you need to run buildbot and have it re-create the entire running environment from scratch
16:02 hmmmm      and then compile a massive piece of software
16:03 hmmmm      and then run ALL of the unit tests for ALL pieces of software on literally 15 different versions of Windows
16:03 hmmmm      the whole process takes about 3 hours and 30 minutes
16:03 est31      hmmmm, this is a stupid GAME not a business software where a failure will cost millions of dollars
16:03 hmmmm      now, there's a reason to ignore a minor change
16:04 hmmmm      oh
16:04 est31      ah
16:04 est31      you only talk about work
16:04 hmmmm      and in the case that somebody else committed another PR in between the time it took for your PR to get re-reviewed and unit tested
16:04 est31      sorry didnt read that
16:04 hmmmm      you need to manually rebase it and then you lose all approval AGAIN
16:04 hmmmm      so you need to go through this entire process before anybody else touches the repository
16:05 hmmmm      i've never wanted to physically harm anything in my life until i had been introduced to this setup
16:05 paramat    oh man
16:05 hmmmm      and then the same people who set this up are telling us that our project moves too slow
16:06 paramat    heh
16:06 est31      :)
16:06 hmmmm      wanna kill something
16:06 hmmmm      oh and you better make sure you have a JIRA issue created for whatever change it is you're making in the appropriate format
16:07 hmmmm      because in order to commit you need to have the JIRA number included in the commit message
16:07 hmmmm      and this JIRA needs to have been approved by the business analyst, and it needs to have been put into the current sprint
16:07 hmmmm      in order to get a specific JIRA story into the sprint you need approval from the manager and scrum master
16:13 Calinou    <est31> hmmmm, this is a stupid GAME not a business software where a failure will cost millions of dollars
16:13 Calinou    and we should keep legacy protocol support? :P
16:14 Calinou    in the non-free world, almost all online games ship with an updater...
16:14 hmmmm      can you devise a reason why legacy protocol support should be dropped?
16:14 Calinou    it's less work, less code (shorter compile times/smaller binaries)
16:14 Calinou    and we can iterate faster
16:14 hmmmm      but there's no reason why compatibility actually requires being broken
16:14 est31      LESS WORK: --> it requires work to remove compat
16:14 Calinou    if you didn't understand it already, games in Linux distribution repositories are intended as showcases, not to be usable
16:14 hmmmm      nobody is making any changes that affect compatibility
16:15 Calinou    (even in rolling-release distros, this can be a problem)
16:15 est31      SHORTER COMPILE TIMES/SMALLER BINARIES: if you take away *one* texture from minetest, you essentially have the same size advantage
16:15 est31      or make it 10 textures
16:17 Calinou    textures don't affect compile times :P
16:17 est31      thats true
16:18 hmmmm      here it's worth repeating
16:18 hmmmm      Jun 04 17:25:43 <hmmmm> the way i change things doesn't necessarily break everything else
16:18 hmmmm      Jun 04 17:25:55 <hmmmm> for some reason it seems like people assume or even want compatibility to be broken between changes
16:18 hmmmm      Jun 04 17:26:10 <hmmmm> there is no correlation between the significance of changes and reverse compatibility
16:18 hmmmm      Jun 04 17:26:58 <hmmmm> people are like cheering on changes that break compatibility for the network because they think that must mean there is a big change and a big change is always for the better, right?
16:18 hmmmm      Jun 04 17:27:25 <hmmmm> and breaking compatibility does not make you more progressive and open to new ideas
16:18 est31      I rather have somebody spend more time on a format that's solid because once its there its hard to remove, than somebody spend less time on multiple bad formats but "iterating faster"
16:19 est31      with the excuse "we can re-design it whenever we want to"
16:19 est31      <Calinou> if you didn't understand it already, games in Linux distribution repositories are intended as showcases, not to be usable
16:19 est31      I agree this is a problem
16:20 est31      but even then its the choice of the distros to do it
16:20 est31      we are an open source project, any distro may take our game and put it in their archives
16:25 xunto      haven't reached 1.0.0 yet @ is caring about compatibility
16:25 xunto      funny)
16:27 est31      semver is just some stupid file on the internet
16:27 est31      what matters is if people use the thing
16:32 paramat    mtgame devs, nore sfan5 game#1124 needed soon so might merge later
16:32 ShadowBot  https://github.com/minetest/minetest_game/issues/1124 -- Mapgen: Add biome fields for riverbed node and depth by paramat
16:50 celeron55  xunto: current minetest can generally load worlds that were created like less than half a year after the beginning of the entire project, and we are proud of it
16:51 hmmmm      but the thing is, this discussion is STUPID because there is simply no actual requirement for compatibility to be broken in order to further develop minetest
16:51 hmmmm      there is no superior thing that can't be done without breaking things from the past
16:52 hmmmm      there is an implicit suggestion that breaking compatibility will enhance minetest in some way
16:52 hmmmm      this is totally wrong
16:56 celeron55  i almost would like to force minetest to start using major version 100 or something so that these trolls couldn't say "haven't reached 1.0.0"
16:57 xunto      >these trolls
16:57 xunto      Sorry)
16:58 sfan5      paramat: i have no idea about mapgen so that pull looks fine to me
16:58 xunto      It was accidential trolling
16:59 celeron55  well i don't actually care; it's just a thing you get really bored of and that stops becoming funny at all
17:00 celeron55  s/becoming/being/
17:00 celeron55  make better jokes
17:00 xunto      Next time
17:00 est31      <hmmmm> but the thing is, this discussion is STUPID because there is simply no actual requirement for compatibility to be broken in order to further develop minetest
17:00 celeron55  that's a promise, then!
17:00 est31      +1 to that
17:02 celeron55  that hmmmm's work PR thing sounds absolutely ridiculous though; it wouldn't take long for me to either change it or show myself the door
17:02 hmmmm      ??
17:02 rubenwardy <hmmmm> and then run ALL of the unit tests for ALL pieces of software on literally 15 different versions of Windows
17:03 hmmmm      ohh
17:03 hmmmm      i forgot i was talking about that even
17:03 est31      shrug if its automated why not
17:03 sfan5      <hmmmm> the whole process takes about 3 hours and 30 minutes
17:03 sfan5      "why not"
17:04 est31      oh
17:04 paramat    ok
17:04 hmmmm      celeron, I stick around because i enjoy the subject matter of what i work on and it's a total work-at-home thing, aside from having to go on week-long business trips a few times a year
17:04 hmmmm      the bureaucracy is overwhelming though and i don't feel bad about not being very productive either
17:05 hmmmm      i have less say on that project than literally any project i've worked on in the past...
17:05 celeron55  i always feel bad about not being productive, no matter the cause, and consider it to be a benefit
17:08 hmmmm      it's a benefit when the productivity is for yourself; not so a faceless, soulless corporation can make an extra buck off of your toil
17:08 celeron55  i don't work for faceless, soulless corporations
17:08 hmmmm      i need to in order to buy food to live
17:09 hmmmm      fwiw I have plans to retire in 10 years
17:09 hmmmm      not going to live my entire life working
17:10 Krock      work work work until you DIE with hopeless and totally DESTROYED </dramatical sound>
17:10 celeron55  i work for companies that appreciate me ensuing my own productivity; managed to find one just lately
17:12 celeron55  it's actually not that great to not have any work at all either
17:13 hmmmm      depending on how i feel i could consider minetest work
17:32 hmmmm      paramat:  i was planning a bit more about the mapgen config change
17:33 hmmmm      starting to wonder a bit if it's really a good idea or not from an interface point-of-view
17:34 hmmmm      first i need to define all of the mapgen defaults in defaultsettings.cpp
17:34 hmmmm      lots of work... eww
17:34 paramat    work on other stuff for now then?
17:34 hmmmm      then i need to add some new Settings methods to copy some settings that match a certain prefix or pattern to a new Settings object, namely ones that begin with mg_
17:35 hmmmm      at this point we have a Settings that is a copy of g_settings that has nothing but settings that start with mg_, plus seed/blocksize/water_level/etc.
17:35 hmmmm      now map_meta.txt gets loaded into a new Settings object
17:36 hmmmm      then i write a new method that merges one Settings into another, overwriting pre-existing entries
17:36 hmmmm      i also am required to maintain a list of settings that were present in the map_meta.txt when it was loaded
17:36 hmmmm      at this point i pass off the Settings to lua to get modified
17:37 hmmmm      you'll have a minetest.set_mapgen_setting("mg_caves_rarity", "some_val_here", true);  for example
17:37 hmmmm      it's like setting a regular parameter except it has a third bool parameter, to override the map_meta.txt settings or not
17:38 hmmmm      if false, which is the default, it'll first check to see if the setting being set is present in the list of setting names we loaded from map_meta.txt
17:38 hmmmm      if so then it does nothing
17:38 hmmmm      if it is not present then it overwrites the setting
17:38 hmmmm      if overwrite = true then it'll just overwrite the setting regardless of whether or not it's one of the settings that had been loaded from map_meta.txt
17:39 hmmmm      so now at this point we have a Settings object that has defaults from defaultsettings.cpp, loaded from the main config file, overrides from map_meta.txt loaded, and modifications from lua done
17:39 hmmmm      the Settings object gets passed along to the mapgens/cavegens/dungeongens/biomegens/whatever
17:39 hmmmm      and in the mapgen you just do
17:40 hmmmm      mgsettings->getFloat("mg_cave_rarity");
17:40 hmmmm      ya dig?
17:41 paramat    i follow
17:41 hmmmm      now we can blow up all the existing "set or get a specific mapgen parameter" lua apis, blow up all the MapgenParams/DungeonParams/CaveParams/BiomeParams structs, and their associated factory methods
17:41 hmmmm      blow up a significant portion of EmergeManager code
17:41 hmmmm      lots of code removed
17:42 paramat    it still sounds good to me
17:43 hmmmm      in order to utilize any of these interfaces, though, it'll require a dependency on Settings
17:43 hmmmm      i wonder if that's a moot point or not, seeing as how deeply entrenched the mapgen code is on the rest of minetest already
17:46 hmmmm      in a perfect world i'd be able to take mapgen_v7.cpp and drop it into another project along with noise.cpp/h and have it work
17:46 hmmmm      :/
17:55 paramat    seems moot to me
18:00 Fixer      OldCoder will bring tsundere.fi world up some day :}
18:13 OldCoder   Not long Fixer
18:13 OldCoder   Not someday; hopefully, this month
18:14 paramat    merging game#1124
18:14 ShadowBot  https://github.com/minetest/minetest_game/issues/1124 -- Mapgen: Add biome fields for riverbed node and depth by paramat
18:18 paramat    merged
18:20 paramat    wow i'm at 300 commits total now
18:23 paramat    Calinou https://github.com/minetest/minetest/issues/826#issuecomment-223819746
18:23 Calinou    maybe it's not there anymore
18:23 Calinou    "In third person mode there's a very short and very small 'jump' of the player model when flying up from touching the ground. I suggest a close."
18:23 Calinou    yeah that might be it
18:23 Fixer      Calinou: i don't get that issue
18:23 Fixer      Calinou: what stairs?
18:24 Calinou    there's a thing in Minetest called stair smoothing
18:24 Calinou    at least in the past, it was enabled even while flying
18:24 Calinou    which introduced a slight (but noticeable) delay when flying upwards
18:24 paramat    est31 shall we close that then?
18:24 Calinou    feel free to close it anyway
18:24 paramat    ok
18:24 Fixer      i have no delay
18:25 paramat    it's certainly hugely improved since my first comment there
18:25 Fixer      it just goes up
18:25 Calinou    Fixer: is there a delay when flying up, compared to flying downwards?
18:25 Calinou    if there is, it's that bug
18:26 Fixer      nope
18:26 Fixer      i'm using PR3810 though
18:26 paramat    closed
18:26 paramat    i tested on master
18:30 Calinou    thanks :p
19:49 Fixer      https://github.com/minetest/minetest/issues/4151 now has backtrace and verbose debug.txt
19:53 hmmmm      what are those backtraces of?
19:53 est31      of a stall on the xanadu server
19:53 est31      essentially the server didnt respond and he did ctrl-c
19:53 est31      then did a backtrace
19:54 est31      so the backtrace *may* include the cause for the stall
19:54 est31      or it may not
19:54 est31      no way to tell from a simple backtrace
19:54 hmmmm      it's the server that freezes, not the client?
19:54 est31      yes the server
19:54 hmmmm      okay
19:55 hmmmm      and gdb.txt and gdb_DEbug_3.txt are two different backtraces from this happening two different times?
19:55 est31      idk
19:55 hmmmm      @ Fixer
19:55 est31      maybe one is terminal backlog
19:55 hmmmm      help us fix this
19:55 est31      and the other one is the .log file
19:55 est31      hmmmm, I've looked at the backtrace
19:55 est31      it seems to be inside craft recipe getting
19:56 Fixer      hmmmm: sorry, thats backtrace originally posted by tenplus1 from his server that has this elusive server stall problem
19:56 est31      and I've been told that the position where the ABM runs on is a furnace
19:56 hmmmm      which one
19:56 hmmmm      gdb_DEbug_3.txt or gdb.txt?
19:56 est31      gdb_DEbug_3.txt
19:56 Fixer      they are the same it seems
19:56 est31      only looked at that one
19:56 est31      you will see an ABM being invoked
19:56 hmmmm      they seem to be exactly the same
19:57 est31      p = {X = 324, Y = 27, Z = -1994}
19:57 est31      this is where a furnace is according to TenPlus1
19:57 Fixer      yes, they are the same, one file copied by hand, another made with gdb, they look the same, he attached both to be sure
19:57 est31      according to him he used mtgame code with the latest version with ABMs
19:58 est31      (mtgame has been changed to use nodetimers for furnaces since)
19:58 est31      so I've asked him to disable the furnace abm
19:58 est31      and it seemed to have worked for him
19:58 est31      no more stall
19:58 Fixer      est31: we need to wait, it can stall later
19:58 est31      hmmmm, the stall is reproducible if you go to certain areas on the server.
19:59 est31      I've tried to download the map via local map saving and installing the mods, but it didnt stall for me
19:59 est31      most likely because I didnt have the mtgame tenplus1 had
19:59 hmmmm      perhaps not
19:59 est31      so there is the chance it is an engine problem, it may be a mod problem as well
19:59 hmmmm      that being said, does the problem go away if a newer mtagme is installed?
20:00 est31      apparently yes
20:00 hmmmm      to be honest this sounds like a bad mod
20:00 nore       est31: have you tried to move items in the furnace locally?
20:00 hmmmm      it takes up too much time in an ABM
20:00 nore       (That will start the nodetimer again)
20:00 hmmmm      does this show up in mod profiling?
20:00 Fixer      est31: wait, he was running 1day old mtg iirc
20:00 nore       hmmmm: problem is that this code has absolutely no loop
20:00 est31      Fixer, 1 day old mtg has already nodetimers
20:01 Fixer      or 2 days old, i build it yesterday
20:01 est31      Fixer, the backtrace clearly shows that an ABM is used
20:01 Fixer      ok
20:01 est31      AND tenplus1 said that he used the last working version with abms
20:02 nore       est31: if you get nothing with latest mtg, it's because the nodetimer is not running
20:02 nore       moving items in the furnace might get it running again
20:09 est31      https://github.com/minetest/minetest/issues/4151#issuecomment-223834567
20:12 est31      hmmmm, tenplus1 is in #minetest-project you can speak with him in that channel
20:23 * tenplus1 is loitering
20:37 tenplus1   bye all
22:40 paramat    uh finally #4191 (am tired) < hmmmm . if it seems good feel free to merge it
22:40 ShadowBot  https://github.com/minetest/minetest/issues/4191 -- Lua_api.txt: Split long lines. Capitalise 'Biome API'. Minor edits by paramat