Time Nick Message 02:52 Zeno` Warr1024, what happened with #4005? 02:52 ShadowBot https://github.com/minetest/minetest/issues/4005 -- Minetest.apk build failure 02:53 Zeno` Warr1024, did your issue turn out to be something else? 03:11 Warr1024 nope 03:12 Warr1024 I think my experiences were the same as yours 03:12 Warr1024 reverting that commit makes it work, ndk update didn't help me 03:13 Warr1024 my desktop and server issues were unrelated 03:27 Zeno` oh 03:27 Zeno` I just can't see your comments but maybe I was looking in the wrong place 04:19 DI3HARD139 Oh Hi there Zeno 04:40 Zeno` hi :) 05:47 Zeno` 2016-04-21 15:33:11: ERROR[AsyncWorker-1]: serves..minetes.net/list?proto_version_min=13&proto_version_max=27 not found (Timeout was reached) (response code 0) 05:47 Zeno` ======= Backtrace: ========= 05:47 Zeno` heh 05:54 hmmmm wtf is this shit? https://github.com/minetest/minetest/commit/8cf26bb0bbc94de437342c84c7b0854feb8957e8?diff=unified#diff-dbdeeb53518283b68748d0b9222b53c1R106 05:54 hmmmm isn't that a clear violation of the lua code style standards 05:55 Zeno` in what way? 05:57 paramat too much alignment? 05:58 hmmmm the if and the statement on the same line 05:58 paramat oh duh! 05:58 * paramat just woke up 05:59 paramat yes it is 06:00 Zeno` I guess 06:01 sfan5 paramat: btw i approved #4010 06:01 ShadowBot https://github.com/minetest/minetest/issues/4010 -- Textures: Replace menu background fallback dirt_bg.png with empty sky texture by paramat 06:01 paramat ok thanks, will merge sometime 06:02 Zeno` hmmmm, if you want to yell at me for missing that before merging then that's ok 06:02 paramat nore sfan5 game#1029 ? easy 06:02 ShadowBot https://github.com/minetest/minetest_game/issues/1029 -- Add: New Heart Texture by KevDoy 06:02 hmmmm d'oh it's fine 06:02 hmmmm it doesn't matter 06:02 Zeno` heh 06:02 hmmmm so when's the release going to happen 06:02 hmmmm i want to be able to work on some new stuff 06:03 paramat 2 weeks of freeze 06:03 paramat maybe 06:03 paramat freeze this weekend 06:03 sfan5 paramat: game#1029 is good 06:03 ShadowBot https://github.com/minetest/minetest_game/issues/1029 -- Add: New Heart Texture by KevDoy 06:03 paramat ok 06:04 paramat hmmmm one thing we may need your help on is #2759 06:04 ShadowBot https://github.com/minetest/minetest/issues/2759 -- Shadow bug at y = 63 in mgv5/mgv7/mgflat/mgfractal/mgvalleys(/mgwatershed) 06:04 paramat i'm going to tackle it during freeze 06:04 hmmmm oh, that.. 06:04 Zeno` I wish I could track this bug down 06:04 paramat but am mystified by this 06:05 Zeno` it doesn't seem to happen in debug builds though :/ 06:05 hmmmm nah i've seen it with debug builds 06:05 hmmmm i am eternally debug 06:05 Zeno` the one I pasted a bit of before? 06:05 paramat he is de bug man 06:05 Zeno` it happens nearly every time I launch the game 06:06 Zeno` aha! 06:06 hmmmm generating in minetest is way too complex 06:06 Zeno` caught it 06:06 sofar I'm getting abm overruns after my game runs for an hour 06:06 sofar could be a timekeeping bug :( 06:06 hmmmm zeno, lol. 06:07 hmmmm paramat, that lighting bug existed way before my patch 06:07 paramat yeah quite possibly 06:08 Zeno` strange 06:08 hmmmm you know what i think would help with debugging this class of problems 06:08 paramat well we had dark air at y=47 a while ago, which you fixed 06:09 hmmmm what if there was a way to 'slow down' light propogation for a specified chunk 06:09 hmmmm and you can visualize it 06:10 hmmmm perhaps the lighting algorithm could be duplicated and instead of setting light it places translucent nodes of varying colors 06:10 Zeno` it's in log.cpp :3 06:10 paramat interesting 06:11 hmmmm how often does this bug happen anyway? 06:11 hmmmm is it readily reproducible? 06:11 paramat or better differing colours of post effect colour 06:11 paramat it is reproducable 06:11 paramat happens 100% 06:11 paramat with terrain level = 62 06:12 paramat all mapgens except mgv6 06:12 hmmmm i'll have a look i guess 06:12 paramat i will for sure 06:13 paramat this is partly why i said a 2 week freeze, for this bug 06:13 hmmmm so all i do to reproduce is set the surface terrain to offset 62 for surface noise? 06:13 Zeno` this is probably the source of all those invalid characters in the log file as well 06:13 hmmmm and then set scale to 0 06:13 paramat yeah 06:13 Zeno` which never used to cause a crash but now it is... 06:14 paramat make sure to read the whole thread for details 06:14 hmmmm once i readily reproduce it i should be ok 06:15 Zeno` http://dpaste.com/22ZD4MH 06:15 paramat then generate the upper chunk first 06:15 hmmmm wait, which noise gets set to 62 exactly 06:15 hmmmm both terrain base and terrain alt? 06:15 hmmmm or what's going on 06:15 paramat mgv7 base and alt 06:15 Zeno` can crash every single time either by connecting before the server list times out or by closing main window before server list times out 06:16 Zeno` prints a bunch of random characters in the "2016-04-21 16:09:44: ERROR[AsyncWorker-0]: serv..." log message and segfaults 06:16 hmmmm wtf paramat 06:16 hmmmm i thought you said the spawn point selection still works 06:17 paramat it won't if all terrain is > 62 06:17 paramat it looks for sea level to sea level + 32 06:17 paramat but yeah could be more robust 06:18 paramat ok i'll work on it 06:18 Zeno` I think the thread is ending before the log finishes outputting it 06:18 Zeno` god 06:19 Zeno` I don't like this :( 06:19 hmmmm humm 06:19 Zeno` log.cpp line 345 (segfault)... source is httpfetch.cpp line 384 06:19 hmmmm i wonder if the problem will disappear if i stop the deco placement algorithm from spawning in the overtop 06:20 Zeno` in httpfetch.cpp the data and strings are valid but they're not by the time log.cpp line 345 is reached 06:20 paramat good idea 06:22 paramat oh but placealldecos is called for node_min to node_max 06:22 paramat maybe it's dusttopnodes 06:24 paramat remember i allowed dust to be placed near full_node_max.Y if the chunk above is generated 06:25 paramat maybe it's placing dark air as dust 06:27 paramat dust has a fallback value of "air", perhaps it should be "ignore" 06:29 paramat .. as dust = ignore disables the dust sprinkle search 06:29 Zeno` #4012 06:29 ShadowBot https://github.com/minetest/minetest/issues/4012 -- Fetch server list segfault 06:30 hmmmm wait a minute 06:30 hmmmm let's think critically here for a moment 06:30 hmmmm this is clearly not on the border of a mapchunk 06:30 hmmmm the mapchunk top is 47 06:30 hmmmm the overtop for that chunk stops at y = 47 + 16 = 63 06:31 paramat yes 06:31 hmmmm the next chunk starts at 48 06:31 hmmmm and the lbound for the underbottom is 48 - 16 = 32 06:31 hmmmm let's see if decorations are disabled, will everything turn black (my guess is no) 06:34 hmmmm hmm 06:34 hmmmm okay it did turn black so it has nothing to do with the decorations at all 06:35 hmmmm let's disable light spreading and only cast sunrays 06:37 hmmmm indeed it only happens when generating top down 06:38 Zeno` actually the crash is happening when the http requests times out 06:38 hmmmm a demon has cast its shadows of evil across the lands of mgv7... 06:38 Zeno` it's got nothing to do with me closing or connecting 06:41 Zeno` ok, now the bug is in an issue I need to work out why this OS won't connect to the public server list 06:41 Zeno` nothing is blocking it (I can visit it in a browser) and changing the timeout from 5000ms to 500000ms makes no difference (crash still happens and still won't fetch) 06:42 paramat heh 06:42 Zeno` blocker! 06:42 Zeno` I have to disable public server list to be able to play! Blocker, blocker, blocker! 06:43 hmmmm that sounds like a pretty embarassing bug 06:44 Zeno` paramat, it's not really a blocker because there is a workaround (of course) :) 06:45 hmmmm :| 06:45 hmmmm should just feed minetest some laxatives 06:45 Zeno` which brand 06:46 Zeno` although... if servers.minetest.net ever goes down lots of people's clients will be possibly crashing 06:46 hmmmm all of them 06:46 * Zeno` merges laxatives 06:46 hmmmm lactulose, miralax, colase 06:46 hmmmm epsom salt 06:46 hmmmm it's going to be one big poo 06:47 hmmmm but at least it'll stop being blocked up and it'll come out 06:47 hmmmm we should have a yearly con for minetest development discussion 06:47 hmmmm call it Shitfest 06:48 paramat meseCon? 06:50 Zeno` meseruns 06:50 Zeno` lol 06:50 Zeno` I'll meet you at shitfest 06:51 Zeno` the inaugural shitfest at Taco Bell LA 06:51 Zeno` I'll be there for sure! Can't wait 06:52 hmmmm oh this is cool 06:53 hmmmm so i have light spreading disabled right 06:53 hmmmm and then i wander into a snow biome 06:53 hmmmm which has the raised snow cover as the dust node 06:53 hmmmm so all of the snow biome is pure blackness while everything outside the snow biome is nice and sunshiny 06:54 hmmmm i have a mod idea 06:54 hmmmm the dark realm 06:54 jin_xi vantablack snow 06:54 hmmmm you need to complete some kind of objective before the darkness swallows up some kind of area 06:55 VanessaE hmmmm: sounds like the plot from Neverending Story. 06:55 paramat ok my idea about dust top nodes fixes this =O 06:55 hmmmm paramat, it has nothing to do with dust top nodes 06:55 hmmmm it's in propogateSunlight 06:55 paramat ok, but this tests as fixed 06:56 paramat will continue to test 06:56 hmmmm i would prefer to fix the actual problem rather than covering up the symptom 06:56 paramat indeed 06:56 hmmmm http://i.imgur.com/r0MA1Aw.png 06:57 hmmmm creeping darkness 06:57 paramat wow 06:57 Zeno` crikey 06:59 Zeno` I like it 07:00 Zeno` maybe it can be rebranded as a feature 07:00 jin_xi oh my god! Its full of features 07:00 Zeno` lol 07:00 hmmmm i laughed 07:03 paramat setting the dust fallback to "ignore" fixes this, so this is due to "air" being placed as a dust, due to biomes where dust is unspecified 07:04 Zeno` jin_xi, can you recommend a mod I should be using to test your particles? 07:05 paramat in dustTopNodes dust = ignore will disable dust placement 07:06 jin_xi no, just digging particles work 'kind of as expected' for now, something is off with time scaling for spawners. also i fixed the offset thing and cleaned up, let me paste again, ok? 07:08 jin_xi http://paste.ubuntu.com/15960672/ <= drop in replacement for particles.cpp using irrlicht triangle selector for collision detection 07:08 hmmmm very nice paramat 07:09 hmmmm i'm sorry about what i said before, it really is caused by the dust nodes 07:09 hmmmm everything is always getting dusted with air 07:09 paramat there is a reason dust is placed in the overtop, so perhaps the lighting call should now cover the overtop 07:10 hmmmm i don't know. there are a lot of extremely subtle things that have changed since i first wrote it 07:10 hmmmm this might have to do with 1 up 1 down generation 07:10 paramat weird how chatting to you makes my head clear, so thanks 07:10 Zeno` ok thanks jin_xi 07:10 hmmmm it's called rubber duck debugging 07:10 paramat well i tried removing that and it didn't help 07:11 paramat (overgen) 07:11 paramat ok so i'll work on this then 07:11 hmmmm dust is written to the overtop because it won't get another chance to generate when the chunk containing the overtop runs 07:11 hmmmm well, at least, that was the original rationale 07:11 hmmmm but now with 1 down overgen, it's not quite the case 07:12 hmmmm you can safely remove node dusting from the overtop since terrain at the chunk border will have another chance to get dusted when the chunk on top is generated 07:12 paramat well it's a change i made to ensure dust on overgenerated trees 07:12 Zeno` jin_xi, can you paste it at dpaste? 07:12 paramat it's essential 07:12 hmmmm oh, touche 07:13 Zeno` jin_xi, so I don't have to manually remove the line numbers (I can't get it as plain text without creating an ubuntu login and I don't want to do that) 07:13 paramat so i guess the lightng call must now cover this dust 07:13 jin_xi oh sure 07:13 paramat so yeah this was my fault :} 07:13 Zeno` paramat, you're a blocker 07:14 Zeno` :p\ 07:14 paramat beta-blocker 07:14 jin_xi http://dpaste.com/2FGH5DY 07:14 Zeno` thanks jin_xi 07:14 hmmmm proton pump inhibitor 07:30 Zeno` actually 07:30 nrzkt no 07:30 nrzkt i don't know what but: no 07:31 Zeno` maybe #4012 might be an almost blocker now because there is no way to disable public server list downloading since I merged kilbith's PR 07:31 ShadowBot https://github.com/minetest/minetest/issues/4012 -- Fetch server list segfault 07:31 paramat still a very minor issue with snow dust in taiga but i'll look into that, for now i'll push this y = 63 fix today 07:32 Zeno` so if the public server list doesn't work, the game crashes and you can no longer just unclick the public server list checkbox as a workaround 07:34 * Zeno` panics 07:34 VanessaE that's a blocker for sure. 07:34 * nore agrees 07:35 paramat milestoned it 07:35 Zeno` I'm having trouble bisecting it 07:35 Zeno` the bug is still there in the last release! 07:35 Zeno` I'll just bisect the entire history I guess 07:39 Zeno` can anyone block servers.minetest.net at their firewall and try and reproduce? 07:40 Zeno` or make the curl request timeout in some other manner (you can probably decrease the time-out from 5000 to maybe 1 in httpfetch.cpp) 07:42 Zeno` actually the timeout is in convert_json.cpp (change it at line 39... e.g. add fetch_request.timeout = 10; or something might do it) 07:43 Zeno` it's not *at* line 39... you'd have to add fetch_request.timeout = 10; :) That might cause it to be reproducible without firewalling the server URL 07:44 Zeno` by default the timeout is 5000 07:45 nore Zeno`: unplugging the network connection between the host is resolved and the server list is downloaded gives no crash but this error message: 07:46 nore 2016-04-21 09:44:28: ERROR[AsyncWorker-0]: servers.minetest.net/list?proto_version_min=13&proto_version_max=27 not found (Timeout was reached) (response code 200) 07:46 Zeno` also "serverlist_url" is a setting 07:46 Zeno` 200? 07:46 nore is yes 07:46 nore s/is// 07:47 Zeno` the crash for me is actually occurring when it's trying to log that actual message 07:47 nore strange 07:47 Zeno` it can be seen in the backtrace (log.cpp line XXX) 07:47 nore do you get a crash with "couldn't resolve hostname" too? 07:48 Zeno` I have a terrible feeling that it's a race condition 07:48 Zeno` I'll change the hostname to something non-existent. one sec 07:48 nore (unplugging the cable *before* asking the serverlist) 07:49 Zeno` yes it's the same 07:50 Zeno` it seems to be timing out before resolving the URL 07:50 nore crash too? 07:50 Zeno` yes 07:50 nore it's probably a race condition indeed :/ 07:51 Zeno` notreal.invalid/list?proto_version_min=13&proto_version_max=27 07:51 Zeno` and that crashes at log.cpp line 345 07:51 Zeno` the response code is now 0 though 07:51 Zeno` hmm 07:52 nore yes, I get response code 0 too for unknown host 07:52 nore except not a crash... 07:52 Zeno` well, this is crazy 07:52 nore were you able to bisect it finally? 07:52 Zeno` no it'll take a while 07:52 Zeno` hmm 07:53 Zeno` nore, can you do me a favour? build a DEBUG version with JIT *disabled*? 07:53 nore I'll try that 07:53 Zeno` (Lua JIT of course) 07:53 Zeno` thanks 07:55 nore how do you enable debug? 07:55 nore can't find the flag 07:55 Zeno` -DCMAKE_BUILD_TYPE=Debug 07:55 nore thanks 07:55 Zeno` np 07:56 nore compiling... 07:56 Zeno` if it happens with just these conditions it's not a blocker... this is very weird 07:59 nore if I start it now with network cable unplugged I get a crash 07:59 Zeno` I wonder if the Lua JIT is hiding it or the Release build 07:59 Zeno` either way at least it's confirmed 07:59 Zeno` even if it only happens under certain conditions 07:59 Zeno` (the normal conditions I use but not the end-user) 08:01 Zeno` very odd 08:03 nore with LuaJIT enabled I still get a crash 08:03 Zeno` so it only happens with debug builds 08:03 nore it's a race condition 08:03 Zeno` that makes me a bit uneasy though 08:03 Zeno` yeah 08:03 nore with normal build: 08:03 nore 2016-04-21 09:58:20: ERROR[AsyncWorker-1]: minetest.net/list?proto_version_min=13_max=27t ounnot found (Couldn't resolve host nCoul) (response code 00)) 08:03 Zeno` that string is incorrect 08:04 nore it is 08:04 Zeno` my log is full of invalid characters as well 08:04 Zeno` all on that log line 08:04 nore in normal build, the race condition was still there, just less severe 08:04 Zeno` *sigh* 08:04 Zeno` why can't it be an easy bug to find 08:05 nore hm, maybe we should add locks for writing things to the debug streams? 08:05 Zeno` can you add a comment to the report please? 08:05 nore doing that 08:05 Zeno` I have to quickly do something 08:05 Zeno` (just in case you're not here when I get back) 08:05 Zeno` brb (10 minutes or so) 08:05 Zeno` thanks 08:18 paramat #4013 08:18 ShadowBot https://github.com/minetest/minetest/issues/4013 -- Biomes: Make dust fallback 'ignore' to fix y = 63 lighting by paramat 08:19 Zeno` nore, it crashes on release for me also 08:19 Zeno` just less often 08:19 Zeno` maybe just 10% of the time 08:19 nore Zeno`: should I mark the issue as blocker? 08:20 Zeno` I'm not sure. What do you think. The only reason I've probably not noticed it before it because I've always had the "public server list" checkbox unchecked 08:20 nore same 08:20 Zeno` it's pretty bad really 08:21 Zeno` if the server list URL goes down nobody will be able to play 08:21 Zeno` well, maybe or maybe not be able to play 08:21 nore well, the obvious fix I see is to make all streams have a lock to ensure there is no problem with concurrent accesses 08:21 Zeno` How about I mark it as critical? 08:21 paramat i'll merge #4010 and 4013 in an hour or so 08:21 ShadowBot https://github.com/minetest/minetest/issues/4010 -- Textures: Replace menu background fallback dirt_bg.png with empty sky texture by paramat 08:22 nore #4013 08:22 ShadowBot https://github.com/minetest/minetest/issues/4013 -- Biomes: Make dust fallback 'ignore' to fix y = 63 lighting by paramat 08:22 Zeno` nore, yeah that would be the reasonable fix 08:22 nore I wonder how much it would slow mt though 08:22 Zeno` nore, unless it causes performance issues... I haven't looked at the code at all 08:22 Zeno` heh yes 08:23 Zeno` I wonder if making a copy of the string would suffice (probably not) 08:23 nore maybe not verbosestream then, with a warning that it is not thread-safe? 08:23 Zeno` for this release I think that any workaround would be reasonable 08:23 Zeno` (within reason) 08:23 nore or, just a lock around the serverlist download code 08:24 Zeno` that might work 08:24 Zeno` err but will that stop closing the window? I suppose it's only for 5 seconds 08:24 Zeno` it might stop quickly hitting connect or enter as well... worth exploring that option though 08:25 nore hm, or just a lock around errorstream 08:25 nore might be enough 08:25 Zeno` if it's an error then even if it impacts performance it probably doesn't matter 08:25 nore maybe what actually happens is that a buffer is re-allocated 08:26 nore and one of the threads keeps a pointer to the old buffer 08:26 Zeno` I think it is, because using valgrind I get a double-deallocate 08:26 Zeno` well even without valgrind I get it: 08:26 paramat hmmmm so you approve of #4013 ? 08:26 ShadowBot https://github.com/minetest/minetest/issues/4013 -- Biomes: Make dust fallback 'ignore' to fix y = 63 lighting by paramat 08:26 Zeno` *** Error in `../minetest/bin/minetest': double free or corruption (fasttop): 0x00007fa3c8012f90 *** 08:26 Zeno` 2016-04-21 15:33:11: ERROR[AsyncWorker-1]: serves..minetes.net/list?proto_version_min=13&proto_version_max=27 not found (Timeout was reached) (response code 0) 08:27 Zeno` (that's in console) 08:31 Zeno` nore, I tagged it "High Priority" 08:32 nore Zeno`: problem is that a string is used as a buffer inside log.cpp 08:32 Zeno` yes I noticed that 08:32 hmmmm paramat, looks good to me. 08:32 nore if we used a fixed-length buffer instead, we might get corrupted or truncated error messages 08:32 nore but no crash 08:32 nore and it wouldn't affect performance 08:33 Zeno` hmm... that's a possibility yeah 08:34 nore and it wouldn't be a problem is messages > 500 chars (for example) are split over several lines 08:35 nore we might even set the threshhold lower, to something like 200 chars 08:37 nore Zeno`: could you do it? (or do you want me to do that, but I don't have much time right now :/) 08:38 Zeno` I can't do it right now but will let you know tomorrow 08:38 Zeno` we could always add a "disable_server_list" setting lmao 08:39 Zeno` I'm just about to cook dinner and do some chores, but after that I'll be looking at it again 08:39 Zeno` can MT be compiled without CURL support? 08:40 Zeno` ENABLE_CURL 08:40 Zeno` I'll have to check what happens when that's disabled as well I guess 08:40 Zeno` but ultimately the log issue does need fixing 08:46 nore Zeno`: could you try that please? https://github.com/Ekdohibs/minetest/commit/02976161d86dbb8f94a5dff112a4557912bd52ec 08:46 nore should fix logging 08:48 Zeno` testing 08:50 nore (it was quicker to code than I expected ;)) 08:52 Zeno` works for release... compiling debug version 08:52 nore submitting a pr then 08:53 Zeno` yep 08:53 Zeno` works for debug also 08:53 nore #4014 08:53 ShadowBot https://github.com/minetest/minetest/issues/4014 -- Make logging use a fixed-length buffer to avoid race conditions. by Ekdohibs 08:53 Zeno` that *was* quick 08:53 nore yay! :) 08:54 paramat i'm about to merge some stuff shall i merge this too? 08:54 nore paramat: you can 08:54 paramat will do 08:54 Zeno` yep code looks good and it fixes the issue, so merge away :) 08:54 Zeno` thanks nore 08:55 nore np :) 08:55 * nore goes back to his homework 08:55 paramat i'll merge 4010 4013 4014 very soon 08:56 paramat what about #4009 any reviews? 08:56 ShadowBot https://github.com/minetest/minetest/issues/4009 -- Fix timer initialization. by sofar 08:56 nore LGTM 08:57 paramat good, Zeno` is 4009 ok with you? 08:57 Zeno` yes it is 08:58 paramat ok will merge that too 08:58 Zeno` ERROR[AsyncWorker-1]: notreal.invalid/list?proto_version_min=13&proto_version_max=27 not found ( not found (TTimeout was reache) (r (ronse code de 00) 08:58 Zeno` heh... at least it's not crashing 08:58 nore paramat: also, don't forget #3907 before 0.4.14 08:58 ShadowBot https://github.com/minetest/minetest/issues/3907 -- Update credits tab by est31 08:59 paramat i won't 08:59 Zeno` paramat, add me back before you do :P 08:59 nore Zeno`: yep :) but corruption of error messages is not dangerous, as long as we don't get crashes ;) 08:59 paramat yeah 08:59 Zeno` nore, I agree totally 09:00 paramat zeno is already added in credits 09:00 Zeno` yay! 09:00 Zeno` I exist again :) 09:01 Zeno` gotta go... bbl. Thanks again nore 09:01 nore paramat: yes, but wasn't #3907 moving them? 09:01 ShadowBot https://github.com/minetest/minetest/issues/3907 -- Update credits tab by est31 09:02 paramat zeno is in current devs section 09:03 nore ah, indeed 09:03 nore I only saw the line that was removed 09:12 paramat now merging those 4 to engine 09:25 paramat merges complete, lots of milestones closed :] 09:25 nore :) 09:32 nore paramat: could you check #3948 please? 09:32 ShadowBot https://github.com/minetest/minetest/issues/3948 -- Escape more strings: formspecs, item descriptions, infotexts... by Ekdohibs 09:32 nore Zeno`: nrzkt: ^ 09:33 paramat sorry this is beyond my understanding, all i could do is a code style check, not sure that counts as an approval 09:51 Zeno` nore, IMO it should be merged sooner rather than later 09:52 Zeno` escapes were added (fairly) recently, right? 09:52 Zeno` i.e. since the last release? 09:52 Zeno` if so then if \v is to be "reserved" 3948 should be merged before release 09:54 paramat i'll re-add to milestones 09:55 Zeno` nore, based on what I can see and what I just said I strongly think it should be merged before \v starts being used in mods created for a release version of mt 09:55 Zeno` thanks paramat 09:55 paramat looks like it's almost ready anyway 09:55 Zeno` yeah 09:58 Zeno` if hmmmm's questions are answered adequately (they seem to be) then I approve 09:58 paramat hopefully est31 and hmmmm will review it 09:58 Zeno` yep, now that's it's a milestone it should be cool 10:00 nore Zeno`: yes, \v was recent 10:01 Zeno` I thought so... remembered seeing it somewhere. So, yes, I support this PR very much (for the aforementioned reasons) 10:21 kilbith to anyone thinking that single-line conditions and bodies is a violation : http://dev.minetest.net/Lua_code_style_guidelines#Misc 10:22 nore Zeno`: I'm wondering, though... Why are there two async workers trying to download the public serverlist? 10:27 Zeno` twice as fast! 10:28 nore ah, they each download part of it? 10:28 Zeno` actually I'm not sure 10:29 Zeno` I was only looking at the backtrace and didn't notice that two threads were downloading it. Is one launched via lua? 10:29 nore possibly 10:29 nore well, the two thread are async-worker 0 and 1 10:30 Zeno` they're both from get_favorites 10:31 Zeno` I don't get it 10:32 Zeno` thread.cpp line 262 is executed twice for the same thing? 10:45 Zeno` there is certainly two threads 10:46 Zeno` void *Thread::threadProc(void *param) receives two pointers (they're different) that leads to the same execution path 10:57 Zeno` yeah so two workers are definitely started 11:10 Zeno` hang on 11:10 Zeno` init.lua calls itself? 11:12 Zeno` based on http://dpaste.com/0VEES0V of course there are two threads for the same thing 11:12 Zeno` once for the main menu and again for async 11:16 Zeno` "main_menu_script" is not a core setting so it loads init.lua again in the false code section 11:20 Zeno` i.e. I'm betting that dofile(core.get_mainmenu_path() .. DIR_DELIM .. "init.lua") gets called twice 12:14 Zeno` anyone else approve #3979 12:14 ShadowBot https://github.com/minetest/minetest/issues/3979 -- Android settings optimization by MoNTE48 12:15 Zeno` since there seems to be only me, Warr1024 and Wayward_One actually even building android builds can we just get this out of the way? 12:17 Zeno` celeron55, any comments? 12:18 Zeno` The independent review that hmmmm asked for (mainly from Mr Batman) seems to have been done 12:22 celeron55 almost nobody seems to be interested in android; i guess it can be merged 12:23 celeron55 monte48 will do some other android-related PRs after it i think 12:24 Zeno` ok I'll merge 12:25 Zeno` I've tested as well as batman and as far as I can tell it improves things 12:25 Zeno` so *shrug* 12:28 Zeno` at least he's contributing back now 12:29 nrzkt Zeno`: i only build for android to doing releases 12:29 Zeno` nrzkt, yep... I really meant build and test I guess :) 12:30 asl97 Zeno`: https://github.com/minetest/minetest/pull/3979#issuecomment-212896109 12:30 asl97 everyone just forget about my comment about client_mapblock_limit :| 12:30 Zeno` what's that comment mean? 12:30 Zeno` oh 12:31 Zeno` where is that comment 12:31 asl97 https://github.com/minetest/minetest/pull/3979#issuecomment-208237085 12:31 asl97 and a few other setting that i also said in https://github.com/minetest/minetest/pull/3979#issuecomment-208724962 12:32 Zeno` oh man 12:32 asl97 MoNTE48 said he will check those setting later and change the pull request 12:32 asl97 ie: the pull request wasn't ready 12:32 Zeno` well he should have said that >:-] 12:32 Zeno` I hate doing it but I will revert history 12:33 Zeno` it's within the time limit 12:33 Zeno` done 12:33 Zeno` adding comment to the PR 12:36 Zeno` added WIP label 12:36 Zeno` *sigh* 12:48 Krock lol.. that's one of the very rare pulls that were merged before they were done 12:48 Krock there are so many outstanding out there which would be done for a long time 13:05 celeron55 asl97: i know it was forgotten, but it seems nobody wanted to actually do anything about it so i dropped it 13:07 celeron55 apparently it now has proper indication about WIP status 13:24 est31 regarding #4005 last time I checked, building android worked for me 13:24 ShadowBot https://github.com/minetest/minetest/issues/4005 -- Minetest.apk build failure 13:24 est31 but I am rebuilding now afresh 13:34 Zeno` thanks est31 13:34 Zeno` I tried on Ubuntu for about an hour (maybe more)... couldn't work out what was wrong 13:34 Zeno` works fine on F21 & F23 13:34 est31 BUILD SUCCESSFUL 13:34 est31 I am on ubuntu 13:34 est31 15.10 13:34 Zeno` hmm 13:35 est31 which ndk version? 13:35 Zeno` I wonder what we're all doing wrong 13:35 Zeno` umm... r10e I think 13:35 Zeno` (from memory) 13:35 est31 i have r10e 13:35 Zeno` do you have other NDK versions as well? 13:36 est31 I have r9d 13:36 est31 should I try with that 13:36 Zeno` I was thinking maybe it was a $PATH thing with a bin for a latter ndk toolchain in $PATH 13:36 Zeno` but I was testing on a livecd and kind of lost all my history once I gave up 13:37 Zeno` toolchain ~= compiler 13:37 Zeno` it's me, Wayward_One and Warr1024 who all had trouble btw 13:37 Zeno` Warr1024, is using debian 8 I think 14:00 est31 so if its a regression it can be bisected 14:00 est31 if its no regression, then that's bad 14:00 est31 once we have the faulty commit, it can be split into pieces 14:00 est31 to find out which part of the commit causes the problem 14:00 est31 is that a sane thing to suggest Zeno`? 14:02 crazyR hey est31 do you by anychance use the mapper script for making a colors.txt file? 14:02 est31 crazyR, you mean leaftest? 14:02 est31 leaftest requires a colors.txt, it does not provide it 14:03 est31 I dont maintain a colors.txt myself, but VanessaE does. 14:03 est31 and there is a tool to autogenerate colors.txt, somewhere 14:03 crazyR I mean do use this: https://github.com/minetest/minetestmapper/blob/master/autogenerating-colors.txt 14:03 crazyR yeah thats the tool, im having issues with it trying to find some who uses it 14:04 crazyR Il ask VanessaE, thanks :) 14:04 est31 yeah she uses it, np 14:05 nore est31: while you're there, could you review #3948 again please? I have fixed the issues you said last time 14:05 ShadowBot https://github.com/minetest/minetest/issues/3948 -- Escape more strings: formspecs, item descriptions, infotexts... by Ekdohibs 14:06 est31 sorry, I gotta go in 4 minutes 14:06 nore np 14:09 est31 nore, why did you add unescape_string here ?https://github.com/minetest/minetest/pull/3948/files#diff-65f34680878a6bd86f3a59ebc0c06c6dR889 14:10 est31 hmm perhaps the commit is too big to review it this fast. 14:10 nore hm, probably for consistency I'd say 14:11 nore because it should be unescaped like the others 14:12 nore else, you can't have formspec characters in a dropdown... 14:12 est31 ah makes sense 14:12 est31 Zeno`, r9e is a build success for me too. 14:12 est31 r9d* 14:19 nore something I'd like to change is how dropdowns are handled 14:20 nore it should return the index of the text that was clicked, or at least, a key-value system 14:20 nore NOT the string that was displayed 14:34 nrzkt nore could be good for static DD but not for dynamic DD 14:35 nore nrzkt: then a key-value system 14:35 nore you can set the key equal to the string then 15:54 Foghrye4 Hello. I would like to see reviews on a new version of pull request: https://github.com/minetest/minetest/pull/3962 15:57 Zeno` Warr1024, where are your comments regarding the android build issue (perhaps I'm looking in the wrong place) 15:58 Zeno` Foghrye4, that will take a while for me to review 15:59 Foghrye4 OK. Approval will be fine too. 15:59 Zeno` well approval is always good isn't it :) 16:00 Zeno` it's a big change 16:03 Zeno` Foghrye4, at a glance it looks like good work 16:03 Foghrye4 Thank you. 16:05 paramat however we're busy with release for the next week and this is not a priority, we'll have more time after release to look at this 16:06 Zeno` yeah I was just about to say that. It's not even that "we're busy" but that freeze is imminent. So it can't really be merged until after that which will give time for a proper review anyway 16:07 Foghrye4 *sigh*. Okay. Hope you wouldn't forget about it, when time comes. 16:08 sofar that's one of the PR's that looks good for pulling early into 0.4.15 so it can be tested 16:08 Zeno` Foghrye4, this is the first I've seen it. From what I can see the idea is good and the code looks mostly good (just trivial style issues). I won't forget about it 16:09 Zeno` sofar, for sure. Something this big should be merged very soon after release so it gets lots of exposure for testing 16:09 sofar there's a few PRs like that 16:09 Zeno` maybe we could add some kind of label? 16:10 Zeno` Not sure what the label should be though... 16:10 Zeno` hmm 16:10 sofar nah 16:10 nrzkt you can add 0.4.15 milestone for this 16:11 Zeno` well I added two anyway 16:11 sofar that's not really needed 16:11 sofar just do a "merge window" style merge 16:11 sofar where anything that looks reviewed and positive gets merged 16:11 sofar then test like crazy :) 16:12 sofar you might as well mark all non-0.4.14 PR's as 0.4.15 16:12 sofar so it's really just not needed 16:13 Zeno` there must be something we can write to prioritise it for merging after release though 16:13 Zeno` so we don't forget 16:14 sofar it's called "go through open PRs" ;) 16:14 Zeno` pfft 16:16 Zeno` there is a testing label 16:16 Zeno` maybe there should be a "needs testing" one 16:16 Zeno` anything so that we're reminded to look at it 16:16 paramat well there's 'high priority'. i would say farmap is that, the next big thing to do after release 16:17 Zeno` the problem is that "high priority" is so vague (I consider it, for example, to apply only to bugs) 16:17 sofar the itemstackmetadata thing is looking promising 16:18 sofar it's not about being priority I think. I think it's about making clear incremental steps right after the release, instead of at the end of the release process 16:18 Zeno` yes, that's correct 16:24 Zeno` ok I'm adding "merge candidate" if people agree 16:25 Zeno` which can mean that the concept is sound and that the code is sound but maybe it needs more review or testing 16:25 Zeno` anything will do as long as there is *something* there 16:25 Zeno` I think merge candidate kind of sums it up 16:26 Zeno` any objections? 16:26 Zeno` or ideas 16:28 Zeno` sofar? paramat? 16:29 paramat can 'testing' do that? 16:29 paramat maybe not 16:29 Zeno` not really because testing is usually applied by the person who makes the PR 16:29 paramat no objections 16:30 Zeno` ok adding 16:30 Zeno` done 16:31 Zeno` so that can mean: "this should be merged as soon as possible because it's a good idea and needs testing and blah blah" 16:31 Zeno` heh 16:31 sofar sure 16:34 Zeno` maybe it should have been "prioritize" but oh well 16:34 Zeno` at least there is a reminder there now 17:15 Warr1024 Zeno`: I never got around to posting anything about it, since it seemed like there were still things we were going to try. 17:16 Zeno` Warr1024, I must be going crazy. I thought you added a message lol 17:16 Warr1024 In a nutshell, on Debian 8 amd64, I was unable to build it using r10d and r10e ndk, but able to build with either ndk after reverting that one commit. 17:16 Zeno` I am officially insane 17:17 Zeno` can you add that to the issue so that others (e.g. est31) can see? 17:17 Zeno` it really worries me that I've already read your messages when they didn't even exist :( 17:17 * Zeno` googles for a clinic 17:24 Warr1024 Zeno`: must have been something we discussed on IRC 17:24 Warr1024 Zeno`: https://github.com/minetest/minetest/issues/4005#issuecomment-213025745 17:25 Zeno` must have been 17:25 Zeno` it's still a bit disturbing that my insanity has been confirmed 17:26 Warr1024 Maybe you're using an IRC client that looks a lot like github's comments interface... 17:26 Warr1024 I'm on irssi, so not much chance of that here... 17:27 Zeno` I'm using the github irc client 17:27 Zeno` :P (just kidding of course lol) 17:27 Zeno` night all 18:46 celeron55_ somebody sent me an email with bad english (saying they're french) where they ask if they could "work on the version Android de Minetest" 18:46 celeron55_ does there happen to be a person here who is french and who cares about android development enough to find something to work on for this guy? 18:47 sofar nrzkt? 18:48 celeron55_ the email address and the content of this message suggest to me that it's a student with a task to do something to an open source project 18:52 sofar been there :) 18:52 sofar the powertop mailinglist is full of those 19:37 dr4Ke Hi everyone. I can't find out how to access the maximum world size in a lua script (mod). 19:37 dr4Ke I know the C constant MAX_MAP_GENERATION_LIMIT is defined (currently at 31000). 19:37 ShadowNinja celeron55_: I'm working on reworking the Android app ATM, so there may be duplicate work. 19:38 sfan5 why do you need to know that constant in lua 19:38 dr4Ke It seems to be used as map_gen_limit in different places of the C code. But I don't know how to get this value in lua. 19:38 ShadowNinja (Update to API level 23, cleanup, bug fixes, proper adjustment to device theme, etc) 19:38 sfan5 it's also unlikely to change as minetest doesn't support more 32768 nodes in each dir (+ and -) 19:39 sfan5 oh that one 19:39 sfan5 dr4Ke: read "map_generation_limit" from the settings 19:42 dr4Ke My goal is to create an object, at a random place in the world, but within its boundaries. And I'm afraid that some user has a minetest compiled with a value below this one. 19:42 dr4Ke Thanks for the map_generation_limit. I'll look at it. 19:55 dr4Ke sfan5: Thank you very much for the map_generation_limit. It was exactly what I wanted. 20:02 ShadowNinja Can anyone review/test #3994? I'd like to get it merged soon before a have another pull depending on it. 20:02 ShadowBot https://github.com/minetest/minetest/issues/3994 -- Upgrade Android build to Gradle build system by ShadowNinja 20:23 ShadowNinja sfan5, nore, nrzkt: ^ ? 20:34 ShadowNinja A text relocation warning has become an error with Android M. I apparently have to make sure everything's compiled with -fPIC, but I'm apparently still missing something... 20:35 est31 ShadowNinja, you have made the project name settable in settings.gradle; its good for forkers who want to keep the diff small. But forking minetest makes no sense if you don't also change the applicationId 20:36 est31 That's what I've meant in #3994 20:36 ShadowBot https://github.com/minetest/minetest/issues/3994 -- Upgrade Android build to Gradle build system by ShadowNinja 20:36 ShadowNinja est31: There isn't a applicationId call in settings.gradle though, it's only valid in the defaultConfig section in build.gradle. 20:37 ShadowNinja Currently you have to change the name in a lot of places for the Android build anyway. My upcomming patch should improve that a bit, but you're always going to have to at least rename the source directory. 20:38 ShadowNinja In addition to updating the build file with the name and package. 20:38 ShadowNinja And ${applicationId} apparently doesn't work in the package field in AndroidManifest.xml, so that too. 20:39 est31 aha 20:39 est31 well 20:39 est31 okay then 20:40 est31 I gonna test whether it builds, and if it does, I'll +1 it 20:40 est31 still it would be cool if you reverted the pointless \ changes 20:42 est31 Warr1024, you say you can reproduce #4005 , can you test something? 20:42 ShadowBot https://github.com/minetest/minetest/issues/4005 -- Minetest.apk build failure 20:47 Warr1024 sure, what's up? 20:50 est31 I am pretty sure I have a fix 20:50 est31 lemme see 20:50 est31 if the fix does not work i have another strategy at fixing it 20:52 est31 Warr1024, does this patch make it work for you? https://github.com/est31/minetest/commit/384d5e44444fff7fb0092e2bc1dc8fd58418e108 20:52 est31 and if not, then can you go through each commit of this branch: https://github.com/est31/minetest/tree/4005-testing 20:52 est31 and tell me the commit that adds the build error 20:53 ShadowNinja Looks like openssl is the issue... 20:53 nrzkt ShadowNinja, yes 20:54 Warr1024 est31: okay, will take about 20 or 30 min. 20:54 Warr1024 I also can't test that the build output works right now, just whether or not it builds. 21:08 est31 ShadowNinja, can you have a look at #4004 21:08 ShadowBot https://github.com/minetest/minetest/issues/4004 -- CONTRIBUTING: disallow signed git commits by est31 21:08 est31 we rely on no signed git commits and it would be good to have it inside CONTRIBUTING.MD 21:09 Warr1024 est31: failed, same problem, for me. 21:10 est31 okay, then try the commits from https://github.com/est31/minetest/tree/4005-testing and find the one that adds the regression 21:10 Warr1024 alright, could take a little while... 21:11 est31 ShadowNinja, I can't apply the patch for https://github.com/minetest/minetest/pull/3994 21:11 est31 it says space before tab in ident 21:12 ShadowNinja est31: s/if there would be/if there were/ 21:13 ShadowNinja est31: Well, that's no good, but it shouldn't berak the patch. 21:13 ShadowNinja Are you sure it's that? 21:13 est31 hrmm seems it cant really cope with .patch files 21:13 est31 thats bad 21:14 ShadowNinja You could also try `git fetch refs/pull/3994/head:gradle` or whatever the ref is. 21:14 ShadowNinja (followed by `git checkout gradle`) 21:16 ShadowNinja scanelf -qT libs/x86/libminetest.so -> libminetest.so: (memory/data?) [0xF1F480] in (optimized out: previous CAST_encrypt) [0xF1F440] 21:17 ShadowNinja It seems like some of openssl's assembly is incompatible with -fPIC (although I barely know what I'm doing with this). 21:18 est31 hrmm make clean_all fails 21:18 est31 lemme try normal make 21:39 est31 idk its somehow broken 21:40 est31 ShadowNinja, doesnt build for me 21:40 ShadowNinja est31: What's the error? 21:41 est31 Gradle version 2.10 is required. Current version is 2.5. If using the gradle wrapper, try editing the distributionUrl in minetest/build/android/gradle/wrapper/gradle-wrapper.properties to gradle-2.10-all.zip 21:41 est31 thats what make clean_all is failing with 21:41 est31 normal make complains about ssl 21:41 est31 but i want to test that again after clean works 21:42 ShadowNinja est31: Well, er, update gradle to 2.12? 21:43 est31 ShadowNinja, I'm on ubuntu 21:43 est31 15.10 21:44 est31 it should work there, no? 21:44 ShadowNinja est31: Yep, just install Gradle 2.12. 21:44 est31 well I have 2.5 why do i need 2.12 just to execute this thing 21:45 est31 whats so awesome about 2.12 that we cant support 2.5 anymore 21:45 ShadowNinja est31: I'm not sure, try downgrading the android build tools version from 2.0.0, if you don't want to update. 21:46 ShadowNinja 1.5.0 might do. 21:47 est31 okay, second problem: it doesn't use path.cfg anymore 21:48 est31 it complains: java.lang.RuntimeException: SDK location not found. Define location with sdk.dir in the local.properties file or with an ANDROID_HOME environment variable. 21:48 est31 our make based system had this path.cfg file where it read sdk and ndk location 21:48 est31 so its not based on env vars, but on that file 21:48 est31 much more practical 21:48 ShadowNinja est31: It writes that to local.properties. Run `make local.properties`. 21:50 ShadowNinja It should read from local.properties too, but that would probably be a bit more difficult than just executing path.cfg. 21:50 ShadowNinja So for now it's duplicated. 21:50 ShadowNinja I should probably add local.properties as a dependency somwhere if I didn't already. 21:50 est31 i did make clean_all and it complained 21:52 est31 also dont forget to change the com.android.tools.build:gradle version 21:53 ShadowNinja est31: Why? 2.0.0 has incremental compilation support. 21:54 est31 well then dont use system gradle, but gradlew 21:54 est31 it then downloads gradle, just for minetest 21:54 ShadowNinja Eh, or not, looks like that was disabled. 21:55 est31 can it be enabled 21:55 ShadowNinja I don't like the idea of bundling generated files in the source tree. 21:55 ShadowNinja ^ incremental compilation was disabled. 21:56 est31 ok what about wgetting gradlew then 21:58 est31 hmm seems to be not that simple 21:59 est31 Well one of the two alternatives, either change the build tools version to 1.5.0 or add gradlew 22:00 est31 I dont want to be required to install random stuff onto my / 22:00 est31 just in order to compile minetest for android 22:02 est31 btw ShadowNinja making local.properties does not work 22:02 est31 that needs fixing too 22:03 ShadowNinja est31: Yep, just noticed that. 22:03 est31 it creates sdk.dir = and ndk.dir = for me 22:08 est31 do we need this signing config stuff? 22:09 est31 it breaks my build 22:10 ShadowNinja It's only used for the release build. 22:11 est31 still it complains about keystore not found even if I just want to do make clean_all 22:11 ShadowNinja Oh, looks like it might not be optional. 22:11 est31 I have removed it and it works well 22:14 est31 ShadowNinja, other than the s/if there would be/if there were/ you +1 my PR? 22:15 ShadowNinja est31: I've completely removed path.cfg and fixed the signing config (put not pushed yet). 22:15 ShadowNinja est31: Yep. 22:17 ShadowNinja Aaand pushed. 22:18 ShadowNinja I'll drop the gradle build tools version to 1.5 too. 22:18 est31 thanks 22:36 est31 ShadowNinja, next bug: 22:36 est31 /bin/sh: 8: [[: not found 22:36 est31 /bin/sh: 12: Bad substitution 22:36 est31 Makefile:763: recipe for target 'apk' failed 22:37 ShadowNinja est31: Er, you should have [[ as a shell builtin. 22:38 ShadowNinja est31: You could replace it with the deprecated [. 22:38 est31 make calls /bin/sh, if [[ a bashism we need to call /bin/bash manually 22:38 ShadowNinja I'm surprised there's still shells around without that though. 22:39 ShadowNinja Maybe it is a bashism. 22:39 est31 /bin/sh is a symlink to bash but bash goes to sh mode when it detects being used as sh 22:39 ShadowNinja It fixes, eg, the [ "x$foo" == "x" ] type of isEmpty hack. 22:40 est31 err 22:40 est31 its a symlink to dash sorry 22:40 ShadowNinja Oh well, I'll change it to [ ]. 22:40 est31 ShadowNinja, what about this http://stackoverflow.com/a/589300 22:42 ShadowNinja [[ isn't really needed in this case anyway. 22:50 est31 ShadowNinja, is it pushed? 22:52 ShadowNinja est31: Yep. 22:56 est31 it still crashes when I do make clean_all 22:57 est31 but thats no big problem 23:00 est31 okay trying a new make again 23:12 est31 ShadowNinja, it errors now with 23:12 est31 Android NDK: ERROR:jni/Android.mk:ssl: LOCAL_SRC_FILES points to a missing file 23:12 est31 Android NDK: Check that jni/../deps/openssl/libssl.a exists or that its path is correct 23:13 est31 and that file indeed seems to not exist 23:18 est31 running make openssl now 23:18 est31 seems to be my fault 23:30 est31 ShadowNinja, sh fails again for me 23:30 est31 it complains "bad substitution" 23:30 est31 /bin/sh: 12: Bad substitution 23:30 est31 make[1]: *** [apk] Error 2 23:31 est31 Makefile:755: recipe for target 'apk' failed 23:32 est31 i have no idea though what you are doing wrongly here 23:32 est31 it just fails :) 23:33 Fixer ShadowNinja, btw, dat "T"-chat redesign has some bad sides :/ like: coordinates are hidden, bug that people need to scrool the text down 23:34 Fixer damn, now my hexchat has scrolling problem 23:52 Warr1024 !tell est31 commit 45db0547 is the first bad commit in my bisect. 23:52 ShadowBot Warr1024: O.K.