Time  Nick        Message
00:05 PilzAdam    I cant reproduce that on my keyboard
00:05 PilzAdam    it works fine
00:06 PilzAdam    but clearing the setting manually in minetest.conf does reproduce the crash
00:07 PilzAdam    although its not very important when a wrong config file crashes Minetest
00:07 diemartin   ok
00:08 PilzAdam    you can of course file an issue on github, though
00:40 Exio4       there should be some safety checks
00:41 Exio4       let say it should at least crash with an error saying 'fuck you' instead of the assert failing
05:52 celeron55   Exio4: well assert is a safety check, but maybe there should be a safety check that tells the user that the config has an empty value
05:53 celeron55   (or maybe it should just assume it means the key is disabled)
12:26 proller     weather finished and rotted: https://github.com/minetest/minetest/pull/892
12:56 celeron55   how about using kelvins for the groups so 0 is naturally invalid? 8)
12:56 celeron55   (yes, not a good idea)
12:59 proller     noo
12:59 proller     too hard to understand
12:59 Weedy       why?
12:59 Weedy       convert C or F on config load to kelvins
12:59 Weedy       use kelvins in code
12:59 Weedy       never worry about 0
13:00 proller     easier make group 0 allowed, its only one check
13:00 celeron55   proller: there are some odd changes in ABMs in your patchset
13:01 proller     removed something unused
13:01 celeron55   move those ABM changes to a separate pull request
13:01 celeron55   they're not really weather stuff
13:01 celeron55   there are interface changes and at least one thing the effects of which you surely haven't tested
13:12 celeron55   (they aren't even in a separate commit; is this really the amount of quality you expect to be accepted?)
13:18 celeron55   you also get much more people to dare to look at pull requests (and that way get them merged) if the pull requests focus on logical parts in code and it's easy to check that they don't do anything unexpected
14:53 sapier      https://github.com/minetest/minetest/issues/910 seems to be a build issue only
14:57 thexyz      sapier: can you try to build it with VS?
14:58 sapier      if you can tell me where to get vs? ... I'm not very eager to build a second toolchain after working around windows issues for about 2 days to get a working mingw toolchain ;-)
14:58 sapier      windows is such a mess
15:00 thexyz      sapier: http://go.microsoft.com/?linkid=9709949
15:00 thexyz      HAVE FUN
15:00 sapier      couldn't someone do this who is interested in vs builds? or at least has a little bit experience?
15:01 sapier      I had to build some libs myself for mingw I guess this won't be a difference for vs too ... and most likely those libs are the issue why it's not working too
15:01 sapier      at least that's what I think
15:02 thexyz      https://forum.minetest.net/viewtopic.php?id=6727
15:02 thexyz      should ask BlockMen to do it
15:02 thexyz      those libs are the issue? then what've you changed to circumvent it?
15:02 sapier      at least we schould correct the issue title if you want proof it's not a general windows issue I can give screenshots
15:02 sapier      -c
15:03 sapier      I built them to match my toolchain
15:03 thexyz      better gimme working build
15:04 sapier      windows lib/dll/toolchain matching is quite crap most ppl use libs/dlls not maching each other but matching enough to get something compiled and linked
15:04 sapier      ok I'm currently updating libogg freetype and curl to latest after that I'll create a zip
15:05 thexyz      indeed
15:05 thexyz      it's possible I fucked something up in the process
15:06 sapier      it's not very hard to mess those windows toolchains up ... took 2 days to get a working one with all libs same time ... most time I had conflicts about version missmatches
15:07 thexyz      I'm not really sure what do you mean by "not maching each other"
15:08 sapier      libs linked against same lib but different versions of that lib
15:08 thexyz      hm..
15:08 sapier      lib not matching dll ... things like that
15:09 thexyz      I think the only precompiled libs I used in my builds are zlib and maybe luajit
15:09 thexyz      hm, no, I definitely built luajit
15:09 thexyz      well, whatever
15:09 sapier      you build freetype ogg vorbis curl gettext yourself?
15:10 sapier      I haven't used luajit in my build ... 2.0.2 didn't compile
15:11 thexyz      oh, right, I didn't build gettext and freetype
15:12 sapier      latest gettext doesn't build too ... they fixed it in 2012 but for some insane reason that fix isn't in current version
15:13 sapier      I'm about to write a tutorial for mingw ... primary for myself but I guess it'll be usefull for others too
15:15 PilzAdam    hm? isnt there buildbot.sh?
15:16 sapier      lol
15:16 sapier      I guess you know my opinion?
15:17 PilzAdam    sorry, I havent read logs yet
15:17 sapier      btw what's a script good for if you don't know what this script is doing ?
15:17 sapier      a single error or outdated url will break
15:17 PilzAdam    the script is easy to read and understand
15:19 sapier      last time I had a look at this script did dowload lots of bilary libs and dlls ... which may be even source of issue
15:19 sapier      but it's some time since I last had a look at it
16:23 celeron55   is downloading binary libraries hard to understand?
16:23 sapier      IF and ONLY IF those really match not but usually urls aren't quite stable
16:24 celeron55   you just read the name and get it from your preferred location
16:24 sapier      that doesn't work as you most likely wont get that exact version if you just read and download
16:25 celeron55   due to minetest being used on linux, it works with most recent versions of libraries
16:25 celeron55   why do you make it so complicated to yourself 8) it's like people deliberately would make the build process hard
16:25 sapier      yes but most recent version of librarys doesn't seem to be available for windows in general ... windows build seems to be broken for some of those libs we use atm
16:26 sapier      gettext for example
16:26 sapier      freetype
16:26 sapier      too
16:26 celeron55   that's true
16:26 celeron55   and it's actually quite sad
16:26 sapier      we're using visual studio zlib dll for mingw build ... strange things like that
16:27 celeron55   windows development slips to microsoft's tools because open source people don't care to support theirs on windows
16:27 celeron55   and then they probably cry that microsoft is such a monster
16:27 sapier      yes and microsoft doesn't care to support platform independent libraries so we're stuck
16:28 celeron55   but basically, if you use windows, you're screwed anyway, so a few old libraries here and there changes nothing
16:29 sapier      atm I try to build freetype but I'm stuck due to current mingw gcc seems to be a little bit more strict than that version they last fixed :-/
16:30 sapier      still as far as I know we want to support windows for minetest ... don't we? ;-)
16:30 celeron55   yes we do, because we're good
16:30 sapier      lol :-) obviously not good enough ;-)
16:31 celeron55   should we release an opt-in poll with 0.4.8 to let regular users tell us what OS they're using? we surely know all developers use linux or bsd, but we don't know what the distribution is within users
16:32 celeron55   if we at some point knew that 70% of users use linux, we could drop windows
16:32 celeron55   it's not likely though
16:32 sapier      windows is still the most common gaming os so we don't have any chance to drop support if we want minetest to spread
16:32 sapier      but maybe we should decide either to use mingw or vs
16:33 PilzAdam    lets rephrase the poll: "Do we want Minetest to spread in windows?"
16:33 sapier      imho that's not a question I don't want to write a game for 10 ppl
16:33 sapier      no don't start to count the 10 was symbolic only
16:34 PilzAdam    Minetest is actually pretty known in the Open Source Gaming Community
16:34 celeron55   but yeah, minetest is (or could become) probably one of the reasons why some could eventually move from windows to linux, as it works on both and they can get familiar on their old OS first
16:35 celeron55   as is gimp, firefox and whatnot
16:40 Calinou     dropping windows would surely alienate a decent amount of players, exclusivity isn't a good thing unless you're forced to do it
16:41 Calinou     Minetest isn't an AAA game too, so it won't do the thing of "oh, a real game that is exclusive to Linux" to the average gamer
16:41 sapier      I'm against dropping windows support but for consolidating. either by dropping vs/mingw support completely or at least cleanup mingw ... atm the most common mingw build seems to be a mixed build using some vs dlls too
16:42 Calinou     also some people are forced to use windows :(
16:42 celeron55   dropping vs or mingw support lowers the chances of minetest having continuous stream of windows builds like currently it happily does
16:42 VanessaE    I don't care for windows at all, but I have to agree - don't drop support for it.
16:42 celeron55   it is not a situation where anyone wants to be, which is the reason why support for both must exist
16:43 celeron55   i mean, the situation where windows builds are hard to get
16:43 NakedFury   remember without windows there would be no linux or users switching to linux
16:43 celeron55   many open source projects are in that exact situation, and it is sad
16:45 celeron55   NakedFury: without windows the users would be either on linux or (more likely) coming from some other platform
16:45 sapier      I think it'd not be that bad to support win32 if it was cleant up once and continously maintaine
16:47 sapier      binary dll/lib packages tend to be deleted once the new version is available but source codes are way more stable so switching to building all our dependencys on windows too might help
16:48 celeron55   having our own builds of the external libraries would be useful
16:49 celeron55   it could be possible to even have such an unordinary thing as a mingw irrlicht with directx support 8)
16:49 celeron55   if somebody cared to make one once and for all
16:50 sapier      at least for mingw I'll have a try ... maybe if I really don't have anything to do I try vs
16:52 celeron55   ogg is very easy to build on mingw, haven't tried the others
16:53 sapier      yes ogg vorbis and (old version of) gettext work
16:54 sapier      freetype doesn't seem to compile with recent mingw at all
16:54 celeron55   are you planning to build openal by yourself?
16:54 sapier      but ppl report it did work with older versions of mingw
16:54 sapier      I haven't thought about openal by now but I think this might be difficult
16:55 celeron55   probably not hard, using this http://kcat.strangesoft.net/openal.html
16:56 celeron55   minetest's builds currently usually or always use creative's random release of the original or whatever version
16:56 celeron55   it's really shady piece of software to say the least, but does work...
16:56 sapier      I used official openal and didn't have a problem ... luajit doesn't build with current mingw too
17:04 nore        -> Voxel Area Entities (in the TODO)
17:04 nore        any chance it will be done soon?
17:08 ShadowNinja nore: -> Big, long-term goals. Probably not.
17:09 nore        It's a Big, long-term goal for a year...
17:11 ShadowNinja nore: There are more important things to be done before 0.4.8.
17:11 celeron55   i think nobody has even started to try to figure out how it even could be hypothetically done
17:12 nore        of course, the question was for 0.5.0, ot something like that
17:12 nore        celeron55, what part of it is difficult to do?
17:12 celeron55   how could i know, i haven't tried to think about it
17:12 celeron55   if you have spare time, there's some boring bug bashing to do; focus shouldn't be on features at this moment
17:13 celeron55   (reason: 0.4.8)
17:13 nore        about fixing bugs, I don't even know one 50th of the code...
17:13 nore        I only know part of environment.cpp
17:14 nore        since most of my pulls were either Lua pulls, or pulls changing environment.cpp
17:14 nore        btw, will force loading ever be merged?
17:19 celeron55   i think it could be
17:19 celeron55   are there any conflicting pulls around at the moment?
17:20 PilzAdam    I dont want to block it, but wouldnt it be better to first check if the object duplication was fixed?
17:20 celeron55   afaik my patch fixed only a certain case of it
17:21 celeron55   a new way of reproducing it needs to be figured out in order to fix it more
17:21 celeron55   it certainly doesn't happen in the "flying with UFO" case anymore
17:21 celeron55   or does it?
17:36 VanessaE    celeron55: homedecor signs.  Place a dozen or so.  give each one some text.  walk away and return after the mapblock has been unloaded.
17:36 VanessaE    if you get more objects than you have signs, it's probably due to entity dupes.
17:38 celeron55   how do the signs work?
17:38 VanessaE    also something of a blocker bug, not exactly related to entities.  Current HEAD of bas080's bees mod (https://github.com/bas080/bees/commit/8dc4d8181a34b7e0385b4720655d81862d86183f) uses particle spawners.  Whatever he's doing with them is triggering that fps dropout.
17:38 celeron55   technically
17:39 celeron55   can you point me to the source code or if you don't care to explain
17:39 VanessaE    the signs use one entity per sign to display text which is built-up from individual textures via the combine function (I didn't write them).  Pilzadam's signs should be sufficient also since it uses the sme concept.
17:39 celeron55   gimme source
17:39 thexyz      https://github.com/xyzz/minetest-mods/blob/master/signs/init.lua
17:40 VanessaE    source code, https://github.com/PilzAdam/signs or https://github.com/VanessaE/homedecor/blob/master/signs_lib.lua
17:40 VanessaE    or thexyz's link
17:41 thexyz      mine is the shortest
17:41 thexyz      :P
17:41 VanessaE    mine's vis a vis kaeza's is the best :)
17:41 VanessaE    but all three do the same basic thing.
17:43 ShadowNinja VanessaE: I beleive particles ahve some optimization issues.
17:44 VanessaE    ShadowNinja: to put it mildly ;)
17:48 proller     celeron55, what exactly odd in #892 ? anything good and works as expected
17:49 celeron55   i took thexyz's mod, placed a sign, wrote "foobar" on it, walked away to let it unload, came back and there aren't extra objects
17:50 proller     celeron55, making something big and split it to 100 pull requests and maintain these pulls 2-9 months - impossble, better to do nothing
17:50 celeron55   proller: yeah, i expect you to make 100 pull requests and also probably most of them will conflict with something
17:51 celeron55   you need to make it a separate commit at least
17:51 ShadowNinja celeron55: Make a dozen and repeat. You can use //luatransform or similar to make a bunch of signs.
17:51 celeron55   you can't bundle up stuff like you have there; the ABM changes have nothing to do with weather
17:51 celeron55   they're generic features
17:51 VanessaE    I was about to say that one sign may be insufficient to trigger the bug, if it can be at all.
17:51 celeron55   ShadowNinja: why would that have any effect?
17:51 proller     abm changes for more range melting around torch
17:52 VanessaE    celeron55: most servers that have signs have dozens of them around the spawn area
17:52 celeron55   i don't see any reason for many signs to cause other behavior
17:52 VanessaE    they're commonly used for rules and "bulletin boards"
17:52 ShadowNinja celeron55: Because there is a low chance of the bug being triggered.
17:53 NakedFury   Sings are 100% always used in lobby areas of servers or the common spawn ground as a place to write the server rules.
17:54 celeron55   why are you talking like i didn't know how servers use signs, lol
17:54 celeron55   ShadowNinja: the random thing is likely to be exactly same for each sign in a bunch of signs
17:54 celeron55   ShadowNinja: it's impossible for the chance of it to be higher by just increasing the amount of signs
17:55 celeron55   seriously
17:56 celeron55   try this on your own computers and don't expect me to make this bug out
17:56 proller     celeron55, i can split by file before commit, but content_abm.cpp  have changes for abm and weather
17:57 VanessaE    celeron55: because you've said yourself that you don't really play anymore.  How is one supposed to know what you have and have not seen?
17:58 celeron55   VanessaE: i don't know, but i figure it's a pretty basic thing on servers and minetest has had signs since years ago
17:58 VanessaE    yes, but maybe you forgot ;)
17:58 celeron55   in any case i'm not digging this more
17:58 celeron55   i tried with one sign and 5 signs, with just walking away and with restarting the server, near and far
17:58 celeron55   your turn
17:59 celeron55   a verbose log of the bug happening is enough to fix it
17:59 VanessaE    my turn has long since been completed :P  but if it works for you then it's probably fixed.
17:59 celeron55   my guess is it's not
18:00 celeron55   proller: it's hard to discuss about your changes when they're lumped into one big thing
18:00 celeron55   proller: expect to wait longer then until somebody is bored enough to go through the whole thing
18:01 VanessaE    s/the whole thing/that set of features/
18:04 PilzAdam    celeron55, what happens often is this: http://mg.viewskew.com/u/sokomine/m/screenshot-3845000104/
18:04 PilzAdam    objects lying arround stack up, but I have no idea how to reproduce that
18:04 PilzAdam    it seems to happen while the area is loaded, though
18:07 sapier1     pa celeron did a entity duplication fix vanessae tests it at her server I don't know result of that test
18:08 sapier1     maybe that issue is solved
18:08 celeron55   the continuation to that story is that VanessaE is still seeing duplication
18:08 PilzAdam    sapier1, read the logs
18:09 proller     celeron55, now all my changes splited into two things. second here  - 882
18:09 sapier1     sorry
18:10 VanessaE    on my server, there are three main sources of entities:  signs, mobs, and pipeworks.
18:10 VanessaE    (s/mobs/simple mobs/)
18:11 VanessaE    I'm pretty sure these have all been wrung out thoroughly; that none of them produce excess entities, that is.
18:25 celeron55   hopefully someone is able to get a log between there not being duplicate entities and there being duplicate entities
19:23 Sokomine    another point that ought to be sorted out before 0.4.8 is the "cyan wool" behaviour: due to the increasement of possible nodes (more than 4k), worlds created before that do have one node each that can't be properly interacted with and that shows as unkown block in the world
19:24 Sokomine    perhaps a script running on database level once could modify these nodes? or is there a better way?
19:24 PilzAdam    IIRC that was only a client server incompatibility
19:26 Sokomine    when i play the map i copied with sfan5s patched client from recrabs server in singleplayer under current dev, those blocks are still unkown. that's no big surprise: they still carry the same number/id in the database
19:27 PilzAdam    IDs are not saved in the database
19:27 Sokomine    i'm not sure weather running an abm on them would work
19:28 PilzAdam    the item names are saved, otherwise you wouldnt be able to load a map when you changed the mod load order
19:28 Sokomine    the database stores a list of nodenames and ids for each mapblock afaik - and the array containing the information which block's at which position is constructed from those ids
19:37 Sokomine    placing new cyan wool works. perhaps it got a new id. still, i'm afraid the map has to be rewritten once
19:59 Evolykane   Just a test message, ignore it.
19:59 * VanessaE  fails to ignore it.
20:34 VanessaE    Sokomine: a command that searches all loaded blocks in some radius around the player for allegedly-unknown blocks, checks if those blocks are truly unknown, and then re-places them, might be a good idea.
20:34 VanessaE    this should be doable in Lua, however slowly.
20:35 VanessaE    s/ blocks/mapblocks; s/blocks/nodes/g
20:35 VanessaE    aw screw it, you know what I meant.
20:52 Sokomine    hmmm, no. when i place a new cyan wool node, it does work and is visible. only the old ones on the map are no longer accessible. thus abms won't hit them either
20:52 VanessaE    abms work via node names.
20:52 Sokomine    needs to be done on a lower level i'm afraid. at least once per world ought to be enough
20:52 VanessaE    but I was thinking more like a one-time command
20:53 VanessaE    check each node in the map, see if it has something hinky about it, re-place it with the same node.
20:53 VanessaE    that would be slow but it would work.
20:53 Sokomine    yes. one-time command to exchange all occourances of the special id though another one