Time  Nick        Message
00:15 paramat     hmmmmm, i can confirm those lighting bugs appear when LVM is used in mgv6
00:28 paramat     hmmmm v5 now segfaults on startup, stu on forum did a backtrace https://gist.github.com/stujones11/cf40d476c04659b884d5 my guess is v5 overgeneration is breaking spreadlight
00:36 paramat     I'll try to fix
00:38 paramat     i'll try calcLighting(node_min, node_max); for v5 as in v6/7
00:39 hmmmm       aghh
00:39 hmmmm       sorry paramat
00:40 hmmmm       too many things to test
00:40 hmmmm       having a widely used development branch is both a blessing and a curse
00:51 paramat     no problem, leave the v5 segfault to me, however i have no idea on the first problem
00:52 hmmmm       oh noes
00:52 hmmmm       what the heck is this horizontal line of darkness
00:53 hmmmm       you have a very.... erm... zigzaggy piece of terrain there
00:53 hmmmm       is that intentional?
01:00 paramat     no, that's the first bug
01:00 hmmmm       aahh
01:01 paramat     see the issue
01:01 hmmmm       i only changed lighting
01:01 hmmmm       nothing else.. how is that possible
01:01 paramat     second bug is v5 segfaulting, possibly due to my overgeneration
01:01 hmmmm       oh
01:01 hmmmm       I thought you meant the first bug was the zigzaggy terrain in your screenshots
01:02 paramat     yes first bug is the zigzag
01:02 hmmmm       and the horizontal line of darkness
01:02 hmmmm       so there's three bugs here
01:02 hmmmm       in any case, i literally did not modify anything
01:02 hmmmm       could you do a bisect on that one to find the bad commit
01:03 paramat     here's the post about the 2nd bug https://forum.minetest.net/viewtopic.php?f=6&t=10908
01:04 paramat     see https://github.com/minetest/minetest/issues/1727#issuecomment-68627007 for first bug
01:04 hmmmm       yup i am looking at that
01:04 hmmmm       that's 3 bugs, not 2
01:05 paramat     yeah i guess
01:06 paramat     basically since your 'fix almost all light issues', any use of LVM with 'vm:set_lighting({day=0, night=0})' produces those dark lines
01:06 paramat     .. in singlenode or mgv6
01:06 hmmmm       i don't see why that would happen..
01:07 hmmmm       there's too much testing required
01:07 hmmmm       any single lighting change requires testing now 8 variants of the same thing
01:08 hmmmm       i'll take a look at it but i have other things going
01:09 paramat     as for v5 segfaulting, i suspect it needs 'calcLighting(node_min, node_max);' instead of 'calcLighting(node_min - v3s16(0, 1, 0), node_max + v3s16(0, 1, 0));'. i suspect my overgeneration is causing the segfault since v7 doesn't have overgenration (yet)
01:09 hmmmm       yeah, it is
01:09 hmmmm       i thought i had gotten it right.. nope
01:09 paramat     i can test the fix for v5 segfault
01:10 hmmmm       gonna need to either change the calcLighting interface or call them separately as a custom sort of thing
01:11 paramat     still, thanks for the awesome work last night ..
01:11 hmmmm       it's all incremental
01:12 hmmmm       fix one thing, break another
01:12 hmmmm       but you learn something from it
01:13 paramat     yep. i can confirm water and cave lighting at y = 47 is fixed
01:14 hmmmm       so i think after we get things into a happy state we should think of low-maintenance strategies to automatically test things like lighting and whatever
01:16 paramat     actually, those dark lines are both with mods of mine that use overgeneration (riverdev in singlenode mapgen, fracture in v5/6/7 mapgen)
01:17 paramat     guess i'll test using a no-overgen mod
01:17 hmmmm       if you don't do anything odd it works perfectly fine
01:17 hmmmm       it works perfect as i tested it to work perfectly
01:18 hmmmm       it's just your overgen things with problems
01:18 hmmmm       in particular it's the fact that calcLighting assumes the border depth is 16 on all sides
01:20 paramat     tested, dark lines with a no overgen LVM mod in mgv6
01:21 hmmmm       with set lighting or nolight
01:21 hmmmm       ?
01:21 paramat     with vm:set_lighting({day=0, night=0})
01:21 hmmmm       if you try it with nolight and remove set_lighting it'd work fine
01:21 hmmmm       i'll have to see what's up with set_lighting though
01:21 paramat     any mod that adds content to mgv5/6/7 can't use "nolight"
01:22 hmmmm       why not?
01:22 hmmmm       if you use nolight, add content, then calc_lighting you're essentially delaying the lighting calculations after you make your additions
01:22 paramat     ah okay
01:23 paramat     how about mods which don't run their mapgen in certain mapchunks of mgv5/6/7?
01:23 paramat     .. fro speed optimisation
01:23 paramat     *for
01:23 hmmmm       there's no speed optimization in skipping processing certain chunks
01:24 hmmmm       you're just delaying the same amount of work from after makeChunk runs to when the mod calls calc_lighting
01:24 hmmmm       it's literally the same thing being executed
01:24 paramat     sorry i mean..
01:24 hmmmm       (albeit it runs while envlocked if called from the mod)
01:25 paramat     mods that don't run their LVM mapgen loop in certain chunks to leave core terrain unchanged, that is obviously a speed optimisation
01:26 hmmmm       not running the loop is fine
01:26 hmmmm       so you local vm = get_mapgen_object("voxelmanip"); vm:calc_lighting(); vm:write_to_map();
01:27 paramat     oh ... they should skip mapgen loop but still do calc_lighting
01:27 paramat     okay good we have a owrkaround
01:27 paramat     heh
01:32 paramat     i'll work on a fix for the v5 segfault thing so don't bother with that for now
02:38 ShadowNinja Got back and finished registered_ores/registered_decorations, will push soon: http://sprunge.us/CKjB?diff
03:40 hmmmm       paramat, https://github.com/minetest/minetest/commit/bdcf03ae36553f4c7ba48d18ef8d67f05590dd60
04:21 Wuzzy       kaeza: #2047 is okay with me. I have tested it.
04:21 ShadowBot   https://github.com/minetest/minetest/issues/2047 -- Fix off-by-one error in `string:split` implementation. by kaeza
04:43 Brains      Anybody know what is up with #2017?
04:43 ShadowBot   https://github.com/minetest/minetest/issues/2017 -- facedir > 23 causes segfault in client
04:43 hmmmm       i think we were waiting for RealBadAngel tocomment
04:45 * Brains    does wish he had checked github earlier today though...  He finally got around to looking at his problem and turns out the cavalry was already on the way.  "It still looks funny to check for out of bounds data and just reset it to something sane..."
04:59 paramat     i've noticed that when entering a world, the progress bar is very slow at 'item textures...' even with texture-light mods
05:46 hmmmm       erm
05:46 hmmmm       anyway paramat, if you're still having issues with watershed, it's being caused by your overgeneration and the only way to fix this is to override calc_lighting
06:19 hmmmm       paramat:  https://github.com/minetest/minetest/commit/aa883409210d2a4a346c552d7cfe5d523f37071f
06:19 hmmmm       I fixed v5
07:04 paramat     thanks
07:05 paramat     looks like this might be able to cope with various amounts of overgeneration
07:08 hmmmm       yup
07:08 paramat     great
07:08 hmmmm       i'm going to rename ManualMapVoxelManipulator to MMVoxelManip, any objections?
07:09 paramat     not from me
07:09 paramat     i abbreviate like crazy in my mods
07:09 hmmmm       yup, i tried to add a typedef before but C++ won't let me do that to classes
07:13 paramat     i have a branch where i'm attempting 1 up/1 down overgen for mgv7, if i can get it to work can i add that to mgv7?
07:13 kahrl       the typedef should work
07:13 kahrl       I see no reason for it not to
07:14 hmmmm       i don't have any of the error messages saved, but basically it accused me of changing it to a struct in forward declarations
07:14 hmmmm       typedef ManualMapVoxelManipulator MMVM;    and then in a different source file, class MMVM;
07:14 kahrl       MSVC?
07:14 hmmmm       nope
07:14 hmmmm       gcc
07:14 kahrl       weird
07:15 hmmmm       and now my ide froze
07:15 kahrl       what if you do struct MMVM;
07:15 hmmmm       meh
07:17 kahrl       oh or actually, try before the typedef: class ManualMapVoxelManipulator;
07:17 hmmmm       I do the typedef in the same file as the declaration
07:17 hmmmm       map.h
07:17 kahrl       seriously, why are all compilers so broken and distinguish struct and class
07:17 hmmmm       i do this directly after the class definition
07:18 hmmmm       now, how do class typedef work with ctor invocations
07:19 paramat     so finally i have closed issue 1727
07:19 kahrl       hmmmm: just fine. Though, considering how I just said how broken compilers are...
07:20 hmmmm       alright I replaced them back again
07:20 hmmmm       let's see the errors...
07:21 hmmmm       error AT the location of the typedef:    error: conflicting declaration 'typedef class ManualMapVoxelManipulator MMVM'|
07:21 hmmmm       error: 'struct MMVM' has a previous declaration as 'struct MMVM'
07:21 hmmmm       this points at one of the forward declarations "class MMVM"
07:22 kahrl       oh... that's something different
07:23 kahrl       you can forward declare a struct or class but you can't forward declare a typedef
07:23 hmmmm       hah...
07:23 hmmmm       so
07:23 hmmmm       I have to either include the header with the typedef, defeating the purpose of forward declaring, or repeat my typedef in every place i forward declare
07:24 kahrl       yeah, sounds bad
07:24 hmmmm       see..
07:24 hmmmm       i think i'm going to put this one on hold
07:25 hmmmm       it would increase programming enjoyment a lot if i did
07:25 hmmmm       but i don't want to shorten a huge data structure used everywhere in the codebase
07:25 hmmmm       change the name of *
07:26 kahrl       are there any other useful subclasses of VoxelManipulator than ManualMapVoxelManipulator?
07:26 hmmmm       there used to be the MapVoxelManipulator
07:26 kahrl       if not, you could consider merging them into just VoxelManipulator
07:26 hmmmm       but that one was completely useless
07:26 kahrl       or VoxelManip if you prefer
07:26 hmmmm       VoxelManip sounds dumb
07:26 hmmmm       "voxel mah nips" as one modder put it
07:27 kahrl       it sounds hot :P
07:27 kahrl       but there's the LuaVoxelManip already
07:27 hmmmm       that ruined the abbreviation for me
07:27 hmmmm       MMVoxelManip
07:27 hmmmm       MMVManip
07:27 hmmmm       which one do you guys like the best
07:28 paramat     the shorter one of course
07:34 kahrl       well I don't really care, I'm not the one who has to use it every day :P
07:34 kahrl       you could call it M
07:34 hmmmm       haha
12:45 Zeno`       For obvious reasons I'd like to merge #2049 ASAP
12:45 ShadowBot   https://github.com/minetest/minetest/issues/2049 -- Prevent client crashing if an NDT_AIRLIKE node is dropped by Zeno-
12:53 kahrl       Zeno`: lgtm
12:54 Zeno`       Ok, merging now
13:09 kahrl       it's bad that any stable version client can still be crashed this way
13:10 kahrl       just a null pointer access, so no real security implications, but still bad
13:10 kahrl       but our development process doesn't really allow us to make a quick bugfix release
13:14 Zeno`       Yeah :/
13:15 Zeno`       It's probably rare (i.e. to those with, for example, maptools and associated privs) but I'm not sure. Even that is bad enough though
13:16 Zeno`       I suppose some server moderators/admins give air as a joke to players as well
13:18 kahrl       well at least it doesn't happen with air
13:18 kahrl       since air has an inventory_image and a wield_image
13:18 kahrl       but some mods define other airlike nodes
13:18 Zeno`       Oh yeah
13:19 Zeno`       well I was testing with maptools:light_bulb, but there are other maptools nodes that have the same issue (obviously)
13:21 Zeno`       I didn't discover the issue though (CWz and deezl did on my server and deezl's)
13:22 Zeno`       It's not "normal" but I appreciated them reporting it in a less public manner
13:29 kahrl       I guess we could make a branch starting at 0.4.11 and backport all the simple bugfixes to it, then release that as 0.4.12
13:29 kahrl       Zeno`: do you see reason enough to do that?
13:34 deezl       good morning
13:35 Zeno`       kahrl, I do in *general*, not just because of this single bug. The version numbering is a PITA though
13:35 deezl       a curiosity....is there any sort of donation system for helping with MT development? for things such as getting an official tablet client out to the play store?
13:36 deezl       so people can find it, and not keep playing the buildcrap versions?
13:36 kahrl       deezl: well there's http://minetest.net/donate
13:36 deezl       oh cool
13:36 deezl       does it get used much? (guessing not)
13:36 kahrl       the money generally goes towards running the servers, but I suppose you can ping c55 if you want to donate for something else
13:37 Zeno`       I guess the issue, then, is making sure the donation goes towards making Android builds for the play store
13:37 deezl       I will try to ear-mark some funds this payday to donate to the cause then
13:37 Zeno`       If that's what you specifically want to donate for (it's kind of like a bounty)
13:38 Zeno`       What's the cost (apart from time)?
13:39 deezl       maintaining servers (such as minetest.net perhaps?) as kahrl said?
13:39 deezl       or inchra.net?
13:39 Wayward_One Zeno`, $25 USD
13:39 Zeno`       Oh, yeah I figured that much. I meant what's the cost for getting an app onto the playstore
13:39 deezl       oh
13:39 deezl       ^^
13:39 Zeno`       $25USD per application?
13:39 Zeno`       I mean $25 for each release?
13:40 Wayward_One no, one time registration fee
13:40 Zeno`       I think someone should ask celeron55 and sapier. Seems like a good idea to me
13:41 deezl       plus, seems some of those *other* tablet clients seem to lag servers a bit
13:41 deezl       it would be better for all of us
13:44 deezl       lol, sapier, I granted you interact on BigPappa's Carbone, so you can at least escape the spawn building when you go there to test. :D
13:53 celeron55   oh
13:53 celeron55   i already have a google play developer account
13:53 celeron55   i can put it there if you want
13:53 celeron55   (i didn't even remember minetest exists when i made it some weeks ago for another thing)
13:54 celeron55   we need to figure out the signing of the apk before i'll put it there though
13:55 celeron55   i don't exactly feel like building it myself
13:57 celeron55   does there exist any kind of guide on how to even attempt that?
13:57 deezl       not sure at all
13:57 deezl       not a coder myself, I just run a couple servers, and enjoy the game :D
13:58 kahrl       celeron55: doc/README.android
14:02 celeron55   i guess i'll attempt it then
14:51 celeron55   what the hell, why is this android build automatically copying all games to the asset path
14:52 celeron55   i have like million games and some of them don't even work:
14:52 celeron55   [aapt] /home/celeron55/softat/minetest/build/android/assets/Minetest/games/dreambuilder_game/mods/player_textures/textures/player_HeroBrine.png: error: File is case-insensitive equivalent to: /home/celeron55/softat/minetest/build/android/assets/Minetest/games/dreambuilder_game/mods/player_textures/textures/player_Herobrine.png
14:52 celeron55   do i have to have a clone of the minetest repository in order to not cause this or is this configurable somehow
15:02 celeron55   (looks like the makefile is quite easily modifiable)
15:06 celeron55   lol, 370MB debug package
15:06 sfan5       probably because of the irrlicht debug info
15:09 celeron55   i bet nobody has ever tried to install a package of this size on a Jolla
15:10 celeron55   i'm not very hopeful of this working because lately none of the builds have been working on this device
15:11 celeron55   only some earlier ones worked
15:16 celeron55   the first boot-up is taking literally forever
15:16 celeron55   it says it's scanning something and the progress bar never rises from 0, and it's taking 100% CPU
15:17 celeron55   why is it "scanning" anything in the first place, shouldn't it just copy the stuff?
15:19 celeron55   it's been going on for like 8 minutes by now
15:19 sfan5       because it needs to check whether you already got a previous version of the minetest files installed
15:19 sfan5       i think it would be better to have the static assets in a zip file or something
15:20 sfan5       that would require rewrites in some places tho
15:20 celeron55   this is nonsense, checking a hundred files shouldn't be this slow
15:20 sfan5       irrlicht can handle zips as filesystem
15:21 celeron55   is there any way i can speed this up?
15:21 sfan5       i don't know of any
15:21 celeron55   there's no explanation whatsoever for why it would take this long in the first place
15:22 sfan5       if you delete /sdcard/Minetest/ it should ideally only do single stat()
15:22 sfan5       and then extract all stuff
15:22 sfan5       but I'm not sure how sapier implemented this
15:22 celeron55   i'm going to slap him in the face; this cannot be sane in any way whatsoever
15:25 sfan5       celeron55: you could try adding this https://github.com/minetest/minetest/commit/35149a10e3b26425e358cb6f4ae0449a052d30a7 (in case you're compiling from the 0.4.11 tag)
15:27 celeron55   sounds like a pretty good idea (of course i am compiling 0.4.11)
15:36 celeron55   okay, that was way faster
15:36 celeron55   more like 100x
15:38 celeron55   now, if it just would work, it might be good, but it's telling me to look at my debug.txt 8|
15:39 celeron55   wtf, i guess i'll have to delete the files and try again
15:40 celeron55   aand now it won't even attempt to start up
15:41 celeron55   i think it stitll thinks the files should be in place while they aren't
15:41 celeron55   still*
15:42 celeron55   i wonder how does it determine that
15:51 celeron55   yeah this is just fucked up now
15:51 celeron55   i reinstalled it and removed the Minetest folder
15:51 celeron55   and it doesn't even attempt to copy any files now
15:51 celeron55   black screen comes and the program quits
15:54 celeron55   debug.txt appears, and in it it doesn't say anything about copying files; just says "lol none of these files exist" and quits without saying it wants to quit
15:54 celeron55   oh well, i quess this means it wants to quit: "ERROR[main]: No future without mainmenu"
15:55 celeron55   before that, "ERROR[main]: cannot open /home/nemo/android_storage/Minetest/builtin/init.lua: No such file or directory"
16:41 CraigyDavi  Is the last login command supposed to show something like this? 2015-01-04T17:10:27Z
16:41 CraigyDavi  Surely the T and the Z shouldn't be there?
16:42 Calinou     standard timestamp, CraigyDavi
16:42 Calinou     not sure about the Z though
16:43 CraigyDavi  Ok
16:46 PilzAdam    Calinou, Z means UTC+0 AFAIK
16:47 PilzAdam    CraigyDavi, read ISO 8601
16:47 Krock       ShadowNinja, https://github.com/ShadowNinja/minetest/commit/72395fc#diff-e2b656616d911eb8d3605c2ef99f50bbR429 how about "f32 fov_degrees = (player->getPlayerControl().zoom)? m_cache_zoom_fov : m_cache_fov;" ?
17:37 sapier      https://github.com/minetest/minetest/pull/2051 can someone plz review and test this? for me it fixes all those particle memory leaks but I don't know how to test it
17:41 VanessaE    sapier: what about jin_xi's/ShadowNinja's irrlicht particles work?  shouldn't that be dealt with first?
17:42 sapier      celeron55 are you there?
17:42 sapier      well vanessaE if you insist on loosing tons of memory each second while they figure out how to do something else?
17:43 VanessaE    well I don't exactly insist on that :-) but I'm just saying given how close they were to getting those particles working, I'd hate to send all their work into the scrapheap.
17:44 sapier      doesn't look like he's here well I don't know which apk he tried to install but the last official one had major startup issues on android 4+
17:44 VanessaE    now as for testing it...  I think Homedecor modpack's "fake fire" mod uses particle spawners for the smoke
17:44 sapier      might explain why it took literally forever
17:44 VanessaE    so if you set a shitton of those (punch the fire to turn the smoke on/off), you might get the test you want.
17:45 celeron55   sapier: i am here; please tell me how to get the android version working
17:45 celeron55   it's installed, but won't start, and won't even attempt to copy the files
17:45 celeron55   and writes a debug.txt complaining files don't exist
17:45 sapier      ok first which version did you instlall?
17:46 celeron55   0.4.11 with the copy speed patch
17:46 celeron55   without the patch it copied files, but it was so slow i stopped it and removed files and applied the patch
17:46 sapier      did you build yourself?
17:46 celeron55   yes
17:46 jin_xi      i'm all for sapiers particle changes, esp if they fix current memory issues.
17:46 sapier      if you did this yesterday you most likely cloned the broken version
17:47 celeron55   i stopped it after it was copying for 15 minutes
17:47 celeron55   then the menu worked but game didn't have all files
17:47 celeron55   today
17:47 sapier      oh ok so it's correct version
17:47 celeron55   0.4.11 + 35149a10e3b26425e358cb6f4ae0449a052d30a7
17:48 sapier      I never saw not all files beeing there as required files are checked on startup, can you send me your apk?
17:48 celeron55   ?
17:48 celeron55   how does it check the files
17:49 celeron55   i can't run this in a debugger because i only have this Jolla, not a real android device
17:49 sapier      on startup minetest checks each single file on sdcard to be at least same size the one we have in our apk
17:49 sapier      if this ain't true we copy it again
17:49 celeron55   but the files don't exist at all, why doesn't it copy them
17:50 celeron55   how can that even possibly fail
17:50 celeron55   it seems completely impossible
17:50 celeron55   also it seems that 35149a10e3b26425e358cb6f4ae0449a052d30a7 broke it
17:50 sapier      maybe because yolla does use different external storage paths we never had this way?
17:50 sapier      jolla
17:51 celeron55   it supports various android storage paths and by symlinking them to its canonical android storage path
17:51 celeron55   -and
17:51 sapier      well maybe something changed in 35... doesn't work with symlinks (just guessing)
17:51 celeron55   what path does minetest use?
17:52 sapier      android external storage path + "Minetest"
17:52 sapier      first one changes between different Android versions
17:52 celeron55   i'm going to give this one go after rebooting the whole thing
17:53 celeron55   altough, it probably doesn't make much difference to rebooting the android virtual machine which i have done many times
17:53 sapier      android is a pain in the as way to many versions out there behaving different.
17:53 sapier      does jolla provide logcat output?
17:54 celeron55   yes
17:54 sapier      asset copy should write anything that might fail there
17:55 sapier      MinetestAssetCopy is the activity name used for the copying code
17:58 sapier      Environment.getExternalStorageDirectory().getAbsolutePath() + "/" + "Minetest" is used as storage directory
17:59 sapier      according to the logs "/home/nemo/android_storage/" should be what Environment.getExternalStorageDirectory().getAbsolutePath() returns for you
18:00 celeron55   it's totally able to write the debug.txt there
18:00 sapier      do you see anything about MinetestAssetCopy in logcat?
18:00 celeron55   for the record, here's one line from debug.txt: 19:59:53: ERROR[main]: FontEngine: failed to load freetype font: /home/nemo/android_storage/Minetest/fonts/liberationsans.ttf
18:01 celeron55   so, minetest's C++ part sees the directory as that
18:01 sapier      so there's most likely something within the java part running before
18:04 celeron55   http://fpaste.org/165897/20481000/
18:04 celeron55   what is filelist.txt
18:04 sapier      prior speedup patch files within apk have been checked by android asset browsing functions
18:05 sapier      those functions are horribly slow and have been cause for initial scan beeing that slow
18:05 sapier      I added a filelist generated on build to workaround that android issue
18:05 sapier      without this filelist nothing is gonna be copied
18:06 celeron55   obviously it's missing for whatever reason (probably has to do with me having to delete all the random broken games it tried to include)
18:06 celeron55   how do i regenerate it
18:06 sapier      make clean_assets
18:08 sapier      but that's gonna cause ALL the games beeing copied again
18:08 celeron55   jni/../jni/src/android_version.h:7:47: error: 'STR' was not declared in this scope #define CMAKE_VERSION_STRING STR(VERSION_MAJOR)"."STR(VERSION_MINOR)"."STR(VERSION_PATCH)
18:08 celeron55   uhm
18:08 celeron55   well this is new, why did this happen now
18:08 sapier      I don't know
18:09 sapier      in worst case do a clean_all
18:09 celeron55   what even is STR, who is supposed to provide it
18:10 sapier      android_version.h should provide it
18:10 exio4       converting to string something, probably \o/
18:10 sapier      it's supposed to be the second line in that file
18:11 sapier      android_version.h is generated by android makefile in Line 699++
18:12 sapier      exio is right STR converts a number define to a string define ... crazy stuff but it's required in order to get c/c++ preprocessor to do this
18:12 celeron55   nah, i already modified the makefile locally to not include other games than the one i want (minetest_game)
18:13 celeron55   (to your previous concern)
18:13 sapier      I guess I should provide a positive list to only include selected games and not all games beeing there
18:13 celeron55   now goes make clean_all
18:13 celeron55   let's hope this fixes the mess
18:14 sapier      well at least you have a monster of a machine to compile ;-)
18:16 celeron55   this build process could use more concurrency
18:16 celeron55   it's wasting time copying files while the CPU does nothing
18:16 celeron55   when building irrlicht and stuff one after another
18:17 celeron55   (one thread using 100% CPU equals doing practically nothing, too :P)
18:17 sapier      I tried to do as much as possible in parallel but it's quite hard to pass the parallelizm flags through various make systems
18:18 celeron55   anyway, copying this resulting madness on my phone now
18:18 celeron55   takes a while over wifi
18:18 sapier      at least the libs and irrlicht should be built with as many jobs as you've got cores
18:19 sapier      but I may have missed somthing of course so if you se some error where flags get lost tell me or just fix it ;-)
18:19 celeron55   yes they are
18:20 sapier      makefile does autodetect number of cores maybe there's a bug
18:21 celeron55   is there any benefit in using the debug apk over the release one
18:21 celeron55   because it's huge
18:22 sapier      only if you can debug on device
18:22 celeron55   it's now copying files
18:22 celeron55   it looks very funny because the layout breaks with long paths
18:22 celeron55   the progress bar is bouncing around like crazy
18:22 sapier      yeah that's an open issue
18:23 sapier      maybe I'm gonna add some "path shortening" feature
18:23 celeron55   wait what
18:23 celeron55   this is copying the whole generated doxygen documentation
18:23 celeron55   holy shit
18:23 sapier      lol
18:23 sapier      you may have missed something to exclude
18:24 celeron55   maybe you aren't supposed to copy the whole doc directory as-is
18:24 sapier      well doc folder is part of assets if your doxygen docu is stored there
18:24 celeron55   well that's where "make doc" puts them
18:25 celeron55   so the android build should support them being there
18:25 sapier      of course we can do finetuning what to include, i never had a doxygen doc there so I didn't realize that issue by now
18:25 kaeza       sapier, does your patch fix that horrible performance degradation when using PSs? (e.g. try the weather mod by Jeija)
18:26 sapier      kaeza I didn't even know about that issue, so if it helps it's gonna help only accidentally
18:26 kaeza       i.e. as a side effect of the mem leaks
18:26 sapier      sorry if you wanna know you're gonna have to try
18:27 sapier      celeron parallelism works for me (mostly) about 70% of total compile time all my cores are at 100% ... but I've got only 4 of them
18:27 celeron55   there are about twenty thousand doxygen files and two thousand git object files (of minetest_game) in filelist.txt
18:28 sapier      git object files?
18:28 sapier      grrr thought those have not been there
18:28 celeron55   games/minetest_game/.git
18:28 sapier      ok ok i see there's still some work left
18:28 celeron55   copying these on the device is about 25% done so far
18:29 celeron55   so... 15 minutes to go
18:29 celeron55   well, this is stupid, i'll stop it
18:29 sapier      are those git object files really in your apk?
18:29 celeron55   i'm not sure
18:29 sfan5       sapier: do you expect /bin/cp to ignore .git files?
18:30 celeron55   stopped it now, i'll make it not copy the doxygen generated documentation
18:30 sapier      I'll fix the find command generating filellist.txt
18:30 sapier      no but apk build ignores .* files
18:30 sapier      so most likely they're in filelist.txt only ... well I'll fix it
18:31 celeron55   itt seems to be saying it's ignoring .gitignore files, but it never says it ignores any .git directories
18:32 sapier      give me a few minutes it shouldn't be a big deal to fix this
18:32 sfan5       sapier: what do you think about making minetest read the static files directly from the apk (or a dir. where android extracts assets)?
18:32 sfan5       that would completly eliminate the need of copying anything
18:33 sapier      if you have any idea how to do this do it but I'd guess it's even more slow
18:33 sapier      you'd have to patch irrlicht in order to get this done
18:33 celeron55   i don't think that's needed for now; we don't want parallel systems to maintain
18:33 sfan5       in case android extracts assets somewhere it will be 0% slower
18:33 celeron55   too much work for little benefit
18:33 sapier      it doesn't
18:34 sapier      actually android never extracts them they're compressed and on each access extracted temporary only
18:34 celeron55   sfan5: it's a server-side thing in minetest in any case; irrlicht can't be used there
18:35 celeron55   the server could of course access the android assets directly
18:35 sfan5       this would require wrapping all C and C++ methods of accessing files
18:35 celeron55   they're just file handles with some weird behavior (like being able to read after the end into the next asset if you don't keep track of the length)
18:35 sfan5       which is ... a bit much work
18:38 celeron55   anyway, like i said, not worth the effort for now
18:38 sfan5       correct
18:39 celeron55   (wrapping file access functions is not hard though; you can override them at the linker level)
18:39 celeron55   (if that's even needed; android might open them by just giving it the right path)
18:39 sfan5       and who'll fix that if it breaks on android <version> for some reason?
18:41 sapier      find . -type d -path "*.git" -exec rm -rf {} \; ;                          \
18:41 sapier      add this line at about 643 where the other exec rm -rf lines are
18:41 sapier      should remove .git folders from apk completely
18:42 celeron55   i removed doc from assets and am now testing with that
18:42 sapier      well upper line is a fix for .git folders not for doc
18:43 celeron55   lol i get a null pointer exception when trying to install the release-unsigned version
18:43 celeron55   i guess the installer can't cope with no key
18:43 sapier      well release unsigned tztz
18:44 sapier      still 0-pointer on this case is stupid they should handle security relevant issues better then that
18:46 sapier      celeron why don't you sign it with your developer key you'd have to do that anyway to get it to play store?
18:46 celeron55   i'm lazy
18:46 celeron55   anyway, the debug build isn't fuckhuge anymore as i removed the doxygen crap from it
18:47 sapier      ok
18:47 sfan5       doxygen docs are 350MB?
18:47 celeron55   ...what, why is the main menu scaling completely different now
18:47 sapier      define "different"?
18:47 celeron55   previously it filled the screen better
18:47 celeron55   now it's tiny
18:48 sapier      it uses your device dpi setting you can change the gui scaling factor in settings
18:48 sapier      I'm working on the font size issue
18:48 sapier      that's not been adapted since the font scaling fixes
18:48 celeron55   lol i accidentally scaled it so big that i can't scale it back
18:49 sapier      lol
18:49 celeron55   time for text editing
18:49 celeron55   god damnit this is awful
18:49 sfan5       sounds user-friendly
18:49 sapier      well I'm working on it ;-)
18:49 sapier      right now with little help
18:49 celeron55   oh, it fails to save the configuration
18:50 sapier      anyone willing to join my efforts?
18:50 celeron55   well, i guess this is a feature
18:50 sapier      not really but in your special case it's a feature
18:52 celeron55   the turn sensitivity is way too high and the view distance is too high and the jump button is unresponsive
18:52 celeron55   such a joyful play experience
18:52 Calinou     mouse sensitivity should be in GUI
18:52 celeron55   the only reason why i'm not dying in 2-deep water is that i managed to turn off damage before starting...
18:52 Calinou     view distance should have a lower maximum value, like 150
18:53 twoelk      try picking up small items like apples on a small screen
18:53 sapier      well that's what parameters are for. I haven't found a valid setting matching even my 3 devices
18:53 celeron55   it's 20
18:53 celeron55   and that results in 11 fps
18:53 Calinou     view distances above 150 aren't really reasonable on most hardware
18:53 Calinou     item size was increased thanks to me :P
18:53 sapier      good thing calinou
18:54 Calinou     larger stacks of items (as well as tools) show up larger
18:54 sapier      ok we should collect all those issues and handle them one by one
18:55 Calinou     so, as a recap…
18:55 celeron55   the view distance is the lowest playable one and this thing gets 11fps at BEST
18:55 celeron55   i guess this device simply doesn't have enough power 8)
18:55 Calinou     view distance currently is 35-240, I suggest 40-160 (I personally don't see the point of playing with a view range below 40)
18:56 Mihaly      ahhh! Hello all, and thank you for your great game!
18:56 sapier      well 11fps aren't worst for android devices ;-)
18:57 twoelk      my old-crap-pc often doesn't have more
18:57 sapier      celeron what device do you have?
18:57 celeron55   building is actually relatively possible once you get the player to a suitable place
18:57 Mihaly      just curious about an error msg I get saying it can't (re) write the conf file, it's mobf that gives the server the error mesg. any help? ideas, things I am doing wrong? tried setting r/w perms on the file
18:58 celeron55   http://www.gsmarena.com/jolla_jolla-5457.php
18:58 Mihaly      but no joy
18:58 sapier      Mihaly: what error message exactly?
18:59 Mihaly      Says it cannot write the conf file. hang on, I'll check the error log file, be patient with me.
18:59 sapier      no problem
18:59 Calinou     Mihaly is on Windows 8 by the way.
18:59 sapier      uh ... never tried there but I'd not guess to be that different on filehandling
19:00 PilzAdam    Mihaly, mod problems should be discussed in #minetest
19:00 Mihaly      well, as I mentioned before it did not  help. ------ no, they told me to try /here/
19:01 PilzAdam    Calinou, why limit the max view range? if its unreasonable for some hardware then it gets automatically limited anyway
19:01 Calinou     PilzAdam, server sends to an extent
19:01 Calinou     the autotuner may set it very high, which is never a good thing
19:01 PilzAdam    the sending of blocks is handled server side
19:01 Calinou     should not go above a certain value so that the view range makes sense
19:01 sapier      then we should fix the autotuner ;-)
19:01 Calinou     imagine you have a very good PC
19:01 PilzAdam    it's not affected by the clientt's view range
19:02 Calinou     view range may be set to 400-500
19:02 Calinou     then when you turn around… very low framerate
19:02 sfan5       sapier: minetest is actually very playable on my tablet
19:02 sapier      on mine too
19:02 sapier      you need about 10 fps to be able to play ... not very smooth but you can
19:02 Calinou     my PC has a GTX 570, if I want to keep a good framerate everywhere, the biggest view range I can use is about 160
19:03 Calinou     (good = 50-60 FPS)
19:03 celeron55   once i set the gui scale to something where i actually can hit buttons, this gets a lot more bearable
19:03 Calinou     playable would be 30 FPS, which is the wanted_fps value
19:03 PilzAdam    Calinou, if your framerate goes below wanted_fps then the autotuner is rather quick at adjusting the view range
19:03 PilzAdam    doesn't even take a second in most cases
19:03 Calinou     my framerate almost never goes below wanted_fps
19:03 Calinou     the autotuner works OK
19:04 Calinou     but it will work worse if the selected min/max are poor
19:04 Calinou     so optimize these values!
19:04 PilzAdam    whats your problem then?
19:04 Calinou     I suggested changing the min/max from 35-240 to 40-160
19:04 Calinou     that's all
19:04 twoelk      minetest was playable, on the last tablet I tried, only the rails extending into the air were irritating and fighting needed lots of training
19:04 PilzAdam    how does lowering the max affect the autotuner?
19:04 celeron55   why
19:05 Calinou     I have to go now, bye
19:05 celeron55   if you want higher fps, you can set a higher wanted_fps
19:05 celeron55   it can't be 60 because some systems can't really keep up at 60 no matter how low the view range is
19:05 celeron55   by default
19:06 celeron55   so they'd see nowhere for no benefit
19:07 sapier      twoelk can you tell me what device you used? I heared about that issue every now and then but never found some common property causing it
19:08 twoelk      it was a cheap tablet of a friend I coaxed into trying mt, sorry I cant remember the make, was some noname
19:09 celeron55   okay but anyway, i can now build working copies of android minetest
19:09 celeron55   so i'll probably put one up on the play store soon
19:10 sfan5       are the horrible description in the mainmenu fixes?
19:10 celeron55   now that i do have a developer account
19:10 sfan5       does it still say "Config MODs"
19:10 sfan5       ?
19:11 celeron55   it says "Config mods" and i can see half of the C and 10% of the s because of font and gui scaling 8)
19:14 sapier      hmm let me fix the font size prior you do upload first version ;-)
19:15 celeron55   fix the button size instead
19:15 celeron55   the UI is fucking horrible anyway, a font doesn't matter when it's this bad to use
19:15 twoelk      https://cdn.mediacru.sh/e3933ac1d16b <- two fotos I was able to take of skyrails on an android tablet
19:16 sfan5       skyrails were already fixed, weren't they?
19:16 sapier      twoelk can you give me your tablet specifications?
19:16 Wayward_One speaking of android builds, anyone seen this one? https://forum.minetest.net/viewtopic.php?f=42&t=9389&start=125 I tried it and it works amazingly well, not sure if it'll help any though
19:17 twoelk      not my tablet as stated befor, It is now 300 km away :(
19:18 twoelk      fotos were taken december 13, so if there was a fix after that it might not be relevant anymore
19:21 Krock       there was a fix for it. problem solved
19:22 twoelk      nice
19:23 sapier      krock what's been the issue?
19:23 twoelk      will advise mc-hardcore friend to update his Minetest
19:24 Krock       sapier, I don't remember exactly. the one end of the texture of this drawtype just went up in the sky
19:25 Krock       LemonLake opened an issue for it AFAIk
19:25 sapier      oh so a texture issue .. ok :)
19:29 celeron55   a vertex coordinate issue, to be exact
20:02 celeron55   i wonder if it's acceptable to have the same android version code for different platform-specific patch builds
20:03 celeron55   or if i should rename the whole thing for my releases to be under net.minetest (the domain that i own) and use my own version codes for it
20:03 celeron55   if it's org.minetest, then it can't really have random fixes as new version codes because those aren't always such that they can be accepted upstream right away
20:04 sapier      rename?
20:04 celeron55   i'm going to take the lazy approach and hope that nothing breaks if i release incremental builds with the same version code
20:04 sapier      uargh
20:05 sapier      last time I did this it took half a day to fix every single occurance ;-)
20:06 celeron55   yeah i've done it for a smaller program, it's messy
20:06 celeron55   you have the name written as text, as partial java module paths, as directories, and as partial directories 8)
20:10 celeron55   sapier: have you thought that the android makefile is a bit hazardous
20:11 celeron55   sapier: for example, it doesn't check whether it could cd to ${ROOT}/assets and then removes files in the current directory
20:11 celeron55   it probably should have "&&" after "cd ${ROOT}/assets" instead of ";", for example
20:11 sapier      hmm no didn't see that issue but you're right about it of course
20:12 celeron55   or | exit 1
20:12 celeron55   or some kind of error handling that can't mess up important files
20:12 celeron55   umm... i mean, "|| exit 1" or how does that even work
20:13 sfan5       the makefile should just execute a shell script
20:13 celeron55   well it does; the scripts are inlined in the makefile 8)
20:13 sfan5       :D
20:13 sapier      I didn't want to write 20 small scripts ;-)
20:14 sapier      benefit and curse of gnumake same time
20:17 sapier      ok added safety check to asset copy ... I'm gonna merge all the fixes at once later
20:19 VanessaE    celeron55: how will you deal with the clones that are already on the play store?
20:20 sfan5       VanessaE: ignore the, because they comply with gpl.
20:20 sfan5       s/gpl/lgpl/
20:20 VanessaE    sfan5: I mean in terms of gravitating users towards the REAL minetest
20:20 VanessaE    especially since there's already one package on the Apple store that calls itself so.
20:20 celeron55   well, i'm going to write that it is the official version
20:20 sfan5       ^
20:21 VanessaE    that's a start, but do you really think that people will take notice of that?
20:21 celeron55   the only problem is that our official version can be worse in some aspects than the clones
20:21 sfan5       this minetest version is actually full-features
20:21 sfan5       full-featured*
20:22 sfan5       i think people notice that
20:24 sapier      well celeron55 I hope to get those issues fixed step by step
20:44 celeron55   i went to the sauna and realized i can pick any subgame i want for this without caring about historical compatibility issues
20:45 sapier      true as it's not gonna be compatible to our current beta release anyway
20:49 celeron55   it can't really focus on action at all as that's terrible on a touch screen
20:50 sapier      true touch control is more targeting towards construction
20:53 celeron55   even exploration is probably too much
20:53 celeron55   it's so clumsy to move anywhere
20:53 sapier      guess that's something we should improve
20:54 sapier      do you still have to jump to get a node up? I had a patch by some time doing this automatically
20:54 celeron55   you do need to
20:54 celeron55   also there is no diagonal movement
20:54 celeron55   i often want to press forward and right or left at the same time
20:55 sapier      well diagonal is more hard to implement :/
20:55 sapier      there are ideas but nothing actual done yet
20:55 celeron55   it should be just the area between up and right, doesn't seem terribly hard
20:55 celeron55   and up and left
20:56 celeron55   or, i mean... well, it should be a circular control to begin with
20:56 VanessaE    sapier: I told ya. :)
20:56 VanessaE    buildcrap uses a similar control doesn't it?
20:56 VanessaE    (similar to what c55 asks for)
20:57 celeron55   i think it does
20:58 celeron55   what if we had a mobile-focused game that just works around all these issues by having a relatively flat map, not too rendering-heavy stuff in the world, the "stair constant" tuned so high that the player just glides up hills and so on
20:58 sapier      well a 9 button pad uses way to much screensize on some devices
20:59 sapier      I tend to the dragable point control
20:59 VanessaE    celeron55: you mean like ShadowNinja's creative server.
20:59 VanessaE    (minus the step height)
21:00 VanessaE    sapier: what about a virtual trackball in the corner?
21:00 sapier      that's what I meant
21:00 sapier      I guess
21:01 sapier      still implementation is tricky
21:01 VanessaE    well you're talking of more like a joystick
21:01 sapier      as we don't have any way to do real full range 360° movement we'd have to map it to keys
21:02 VanessaE    I was thinking more like a desktop trackball, i.e. that: http://upload.wikimedia.org/wikipedia/commons/9/96/Trackball-Kensington-ExpertMouse5.jpg
21:02 celeron55   well the main issue here is that having to jump in order to move around is fucking horrifying
21:02 celeron55   fix that and it's 900% better
21:02 celeron55   all else is just extra
21:02 VanessaE    celeron55: I can agree with that - even with normal key/mouse controls it can get annoying depending on the terrain.
21:03 celeron55   well it's not anywhere near that, don't even compare that
21:03 sfan5       also the touch controls suck if you attack a mouse to the android device
21:03 sfan5       but that is probably nothing we have to worry about
21:03 sapier      give me about 20 min and I'll tell you the line where to fix ;-)
21:04 celeron55   how many people are there in this world who connect a mouse to an android device and expect it to work fine? :P
21:04 VanessaE    celeron55: nono, I get what you're saying, I've played on both sapier's build and on buildcraft.  it's ... cumbersome.
21:04 VanessaE    it's just been a while.
21:04 sfan5       the mouse actually worked fine with the system UI, chrome and other apps
21:04 sapier      VanessaE just always wants MORE ;-)
21:04 sfan5       except that right click didn't do anything
21:04 VanessaE    sapier: not in this case :P
21:08 sapier      celeron55 localplayer.cpp L 208 increase the player stepheight by 0.5
21:13 celeron55   doesn't seem to work
21:14 sapier      strange works like charm on desktop
21:14 VanessaE    that reminds me - something was changed in the step height such that if you RUN up a long flight of stairs (say 10 nodes worth), you "sink" into the stairs a bit.
21:15 VanessaE    (on PC, can't speak for android)
21:15 celeron55   well i'm testing it on desktop
21:15 sapier      float player_stepheight = touching_ground ? (BS*1.1) : (BS*0.7);
21:15 celeron55   wait
21:15 sapier      is what I wrote
21:16 celeron55   actually yes, it works as expected
21:16 celeron55   i'm just dumb
21:17 sapier      guess in the real fix I'm gonna make player stepheight configurable
21:17 sapier      this way it's not android only
21:18 celeron55   random thing to do: android should have performance-oriented defaults for rendering, which is, disable 3d clouds and disable fancy leaves
21:18 sapier      do those changes give that much performance?
21:19 sapier      on my devices difference was minor
21:22 celeron55   well, 3d clouds probably don't really matter, but the leaves do if you go in a tree-dense area
21:22 celeron55   they *definitely* do
21:22 celeron55   i personally like 2d clouds much more visually though
21:22 sapier      maybe I didn't have that many trees visible when I tried. guess changing this is a minor issue
21:23 celeron55   so for me, nothing is lost and something is gained by disabling 3d clouds
21:24 sapier      ok I'm gonna change it with my next merge of android changes too
21:37 celeron55   now i know why the turn control is terrible
21:37 sapier      can you be more precise? ;-)
21:38 celeron55   it has a too high threshold; if i raise my finger from the screen, it requires me to turn more than i want when i want to turn a bit again, and if i keep my finger on the screen without moving it, it starts digging random stuff even if i'm trying to just fine-tune my direction when walking around
21:39 celeron55   it shouldn't try to dig when you walk, to begin with
21:39 sapier      threshold is configurable
21:39 celeron55   i don't care, it's wrong!
21:39 sapier      well it's wrong anyway as correct value depends on your device
21:39 celeron55   what setting name
21:39 celeron55   it should be relative to the degrees the player is going to turn
21:39 celeron55   not to screen pixels
21:39 sapier      it should be configurable in settings
21:40 celeron55   pixels change, but what does not change is how much the threshold can be relative to the actual turn speed
21:40 celeron55   turn amount*
21:40 sapier      well if you have an idea how to calculate this?
21:40 celeron55   i'd guess it's linearly calculated somewhere to the other direction
21:41 sapier      I don't understand why this is different to mouse turn control?
21:41 celeron55   it probably isn't
21:42 celeron55   except that mouse doesn't have a threshold value
21:42 sapier      true
21:42 celeron55   anyway, the threshold should be configured in degrees, not pixels or scaled pixels
21:42 celeron55   i guess
21:43 sapier      well it's actually a threshold to be used to compensate screen blur
21:43 celeron55   but more importantly i don't want to dig stuff when i walk!
21:43 celeron55   or equally importantly actually
21:43 sapier      hmm blur is wrong
21:44 celeron55   both of those are very annoying; in creative mode it actually manages to destroy terrain unintentionally
21:44 sapier      you don't unless you keep your finger at exactly same location on screen
21:44 sapier      once you move digging is disable
21:44 sapier      d
21:44 sapier      hmmm
21:44 celeron55   of course i do when i want to move straight ahead
21:44 sapier      maybe we could disable digging if you press one of the move keys too
21:45 sapier      would solve this issue
21:45 celeron55   i want to rest my finger somewhere anyway when i'm walking around, might as well use the screen in the same way as turning but not actually turning
21:46 sapier      if you wanna be able to rest your finger the only always working way is add another button to switch meaning of your finger
21:46 celeron55   because i want to turn tiny angles all the time anyway to keep myself going in a straight line
21:46 celeron55   and then the threshold comes and absolutely hates me
21:46 celeron55   "let's turn 1 degrees" "no, you dig now here in random places" "NO" "okay let's turn 35 degrees" "GOD DAMNIT"
21:46 celeron55   there's how it goes for me
21:47 sapier      hmm
21:47 celeron55   if walking forward, just remove the threshold and never dig
21:47 sapier      any way to fix this I see right now will use another big part of screen
21:47 celeron55   simple as that
21:47 Warr1024    maybe an option to multi-touch to dig?
21:47 celeron55   you can have the large threshold when not walking
21:47 celeron55   it probably makes sense then
21:47 sapier      we're already using multitouch to place
21:48 sapier      maybe short multitouch to place long to dig
21:48 Warr1024    hm, I thought it was double-tap to place...
21:48 sapier      no
21:48 Warr1024    long multitouch could work
21:48 sapier      its target with one finger tap with another finger
21:48 celeron55   sapier: do you now understand how i think this could be easily solved?
21:49 celeron55   if(walking){ turn_threshold=0; allow_dig=false; }
21:49 celeron55   nothing else needed
21:49 sapier      well that's a minor change ... give me 10 min
21:51 celeron55   does someone agree that setting time_speed=0 would make sense for this play store build?
21:52 sapier      time_speed?
21:52 celeron55   that makes singleplayer permanently daytime
21:52 VanessaE    noooooooooo
21:52 celeron55   i don't think almost anyone wants to look at the 10-minute nights on their mobile device; it's almost impossible to do anything
21:53 celeron55   it should be a graphical setting though
21:53 VanessaE    why not adjust the ratio then
21:53 celeron55   it isn't
21:53 celeron55   that isn't even a textual setting
21:53 celeron55   what's so valuable about nights?
21:54 celeron55   they just make it impossible to explore when you want to
21:54 VanessaE    some people like to build structures that only have their best appearance at night
21:54 sfan5       +1 disabling night-time will stop all the people complaining that there are no monsters
21:54 celeron55   on desktop you can see something at night; on a phone or tablet you really can't
21:54 VanessaE    e.g. via artificial lighting
21:54 celeron55   on tablets?
21:54 sfan5       (except they'll complain that there are no animals)
21:54 VanessaE    think Eiffel Tower
21:55 celeron55   holy fuck show me the person who is going to build the eiffel tower on a tablet
21:55 VanessaE    *sigh*
21:56 VanessaE    ok make it always-daytime if you want
21:56 celeron55   i won't play it personally, i'm just trying to guess what people would like
21:56 Amaz        I think always day on mobile would be good.
21:56 celeron55   when i play something on my phone, i want to waste 5 minutes while waiting for something
21:57 celeron55   a nighttime in minetest does not suit that
21:57 sapier      not sure if it's gonna be enough celeron but could you try adding :
21:57 sapier      if (m_move_id != -1)
21:57 sapier      m_move_has_really_moved = true;
21:57 sapier      to touchscreengui.cpp L 699
21:58 sapier      wait L 685 may work better
21:58 celeron55   VanessaE: once there is something more sophisticated in place, i'll change the defaults
21:59 celeron55   but my primary goal is to make this work as smoothly in the general case as possible
21:59 celeron55   with the few hours i'm willing to use for quickly doing that
22:00 celeron55   sapier: note that i'm on 0.4.11, not master
22:00 celeron55   i need to cross-check all these line numbers from github
22:01 sapier      ok it's in TouchScreenGUI::step
22:01 sapier      right below btn->repeatcounter += dtime;
22:02 celeron55   this sounds like a very dirty hack 8)
22:03 sapier      well it's just a first try
22:04 sapier      but it's not that dirty it's using same mechanism switching from dig to look  as used now just from a different location
22:04 sapier      hopefully I didn't miss any additional dependency as I haven't tried yet
22:06 celeron55   now that moving around will probably be bearable, i'm going to look if some game is better at providing stuff to explore than minetest_game
22:06 celeron55   (without being a resource hog)
22:06 celeron55   i'm fine with that being the initial nature of this
22:06 Amaz        Voxelgarden?
22:06 celeron55   that's one option
22:07 celeron55   it doesn't have a terribly lot of content in any way though
22:08 Amaz        BFD? Bit resource heavy, maybe...
22:09 Wayward_One Carbone, perhaps?
22:09 celeron55   too heavy, and even right away the hotbar is too larg
22:09 celeron55   +e
22:10 Wayward_One maybe minetest_game with some extra (possibly voted-on) mods?
22:11 Wayward_One like the android build's own version of minetest_game
22:11 sapier      who's gonna maintain it?
22:11 VanessaE    mt_game along with some of the basics like moreblocks, homedecor, etc. seems like the best way imho
22:11 Wayward_One ^^
22:11 celeron55   i would like it to have a wider array of biomes
22:12 VanessaE    any other game is likely to stress a mobile's RAM limits too much.
22:12 celeron55   simple but varied biomes are relatively cheap
22:12 Amaz        ParagenV7?
22:12 VanessaE    Amaz:  +1
22:12 celeron55   and nobody is doing them very much, except voxelgarden a bit
22:12 VanessaE    (if it can run fast enough)
22:13 Amaz        VanessaE, that would be my worry... It isn't the biome api, so...
22:13 VanessaE    "the biome API" is still totally experimental, per hmmmmm's declaration earlier.
22:13 Amaz        Yeah :P
22:13 VanessaE    so I guess that's out.
22:15 Amaz        Apart from the biome api, paragenv7 is the only mod I know of that provides varied biomes fairly fast. But, I'm not sure if it would be fast enough...
22:17 celeron55   is there something that uses the biome api and is great?
22:17 VanessaE    celeron55: you're not allowed to use the biome API
22:17 celeron55   because i'm not going to care if something does messy map generation borders in mobile minetest
22:17 VanessaE    and the only thing I know that uses it to any real degree is Jordach's BFD.
22:17 Amaz        BFD is the only one I know of.
22:17 celeron55   when the biome api is replaced with something else and the game is updated to use the something else
22:17 Amaz        Or Ethereal.
22:18 celeron55   on a server it matters, but not on the singleplayer game on a phone
22:32 celeron55   does someone have a quick fix for it being almost impossible to dig only one node in creative mode
22:32 celeron55   it's ridiculously spammy with the finicky touch controls
22:32 celeron55   well that wasn't a very descriptive adjective
22:32 celeron55   clumsy rather
22:33 paramat     concerning step smoothing, i still find it too sluggish, try flying up in third-person to see. it was adjusted recently here https://github.com/minetest/minetest/commit/1c9f05d792562374046e74ad3eb75988d529b15c i will try to tune
22:36 paramat     sluggish step smoothing is actually nauseating
22:36 celeron55   i don't think there was anything from with it before
22:36 celeron55   wrong*
22:37 celeron55   (wtf)
22:47 celeron55   ahh, now i'm finally getting this to a playable state in creative mode
22:47 sapier      how? :-)
22:49 celeron55   many commits worth of hacks, that's how
22:49 sapier      lol
22:49 sapier      nothin we can use in future versions?
22:52 celeron55   i'll publish them on my github but only after i publish this apk and everyone says they love it
22:52 celeron55   8)
22:53 sapier      cheater ;-P
22:56 celeron55   i think i'll have to use minetest_game with a few mods
22:57 celeron55   not sure how to organize this, maybe i'll make a new repo to host the engine and the game in one ugly glump
22:58 celeron55   or maybe i'll call the game something else
22:59 celeron55   i guess that would be the "i still respect some of you guys" solution
23:06 celeron55   i'll call the game "straightforward"
23:06 celeron55   that's the general idea anyway, lol
23:17 * twoelk    smells a new android specific subgame
23:22 kilbith     is there a way to close the comments but not a pull request itself ?
23:22 sapier      what's "close a comment"?
23:23 celeron55   i think what i'm going to do is use v5 with a tightened-up paragenv7
23:24 celeron55   this is the best i can do to fill a tablet with glorious stuff
23:24 kilbith     sapier: simply lock the conversation on git
23:24 kilbith     such as on a forum
23:24 sapier      for what reason? a pull is supposed to be discussed about
23:25 sapier      why do you wanna stop ppl from doing (valid) comments?
23:25 kilbith     same reasons like on a forum or youtube, avoid the stupid comments
23:26 paramat     celeron55, paragenv7 needs some essential speed optimisations i recently learnt from hmmmm, could you give me a few hours to update it first?
23:26 sapier      there's a major difference a closed forum thread is not gonna be used for something while a pull most likely is supposed do be merged
23:26 sapier      if you don't want it do be merged close it but if you want it to be merged you can't stop ppl from telling their oppinion
23:27 kilbith     the stupid conversation can influence alot the decision to merge or not
23:27 sapier      well it's not bad style to just stop ppl from discussing if discussion ain't going the way you want it ;)
23:28 kilbith     well, i'm afraid to offend verbally
23:32 paramat     also the biomes are big you might want to reduce temp/humidity spreads
23:38 celeron55   paramat: yeah i already reduced them to 512 and 96
23:38 celeron55   i'm about to test the performance on my phone now
23:39 celeron55   kilbith: you just have to trust that the people who have the power are not stupid enough to let stupid comments lead their decisions
23:40 celeron55   i do
23:41 paramat     there's a mistake that will be eating up loads of memory (2D perlinmap size.z should be omitted), also noise objects should be created only once, so fixing these should help performance on a phone
23:42 paramat     if there's glitches at y=47 i can fix those too
23:44 paramat     BTW the jungletrees have climbable trunks
23:45 kilbith     celeron55: when someone starts to give a new "curve" in a conversation, the devs tend to drift towards the curving to give reasons to let them freeze the PR. this is my subjective feeling and i'm afraid of that
23:49 crazyR      can anyone tell me if it possible to use TAB key as a shortcut to a function within one of my mods?
23:53 celeron55   holy shit i removed homedecor completely and this can barely run for 4 seconds before getting killed by the kernel low memory killer
23:53 celeron55   looks like i have a little under 400MB to use before it's dead
23:54 celeron55   moreblocks is way too much too
23:54 * celeron55 rips everything out
23:55 celeron55   i like constraints though
23:57 sapier      what's your total ram size?
23:57 celeron55   minetest_game+paragenv7 manages to barely stay under that limit
23:57 celeron55   1
23:57 celeron55   now it got killed again
23:58 sapier      1gb?
23:58 celeron55   paramat: how much do you think the excess memory consumption is after generating like 4 chunks?
23:58 sapier      celeron right now we have major memory leaks in minetest I don't know if they've been there in 0.4.11 too
23:58 celeron55   it was fine with bare minetest_game, i just need to somehow fit paragenv7 to this equation
23:59 sapier      ok
23:59 paramat     not sure, but try deleting the z component of 'chulens' line 114