Time Nick Message 00:00 RealBadAngel all of them 00:00 RealBadAngel i am not talkin about master 00:00 ShadowNinja RealBadAngel: Yours should disable compilation seperately. 00:01 RealBadAngel why so? 00:01 PilzAdam ShadowNinja, thats not how his system works 00:01 RealBadAngel introduce extra lags? 00:01 RealBadAngel put there dozens of "ifs" ? 00:02 RealBadAngel pointless 00:02 PilzAdam he puts all the features into one shader 00:03 RealBadAngel if drawtype can have bumpmapping, parallax, waving and whatever all is in one shader file 00:03 RealBadAngel configurable via settings 00:03 RealBadAngel shaders are divided only by drawtype 00:03 RealBadAngel not the effects theyre providing 00:04 * VanessaE is back. yay. 00:04 RealBadAngel yay. thats it. ;) 00:05 RealBadAngel so, ShadowNinja, definitely not. disabling one shader will cause one or more drawtype to be unaviable. 00:06 RealBadAngel in best case, without any effects applied it will miss lighting code 00:06 RealBadAngel and will become black ingame 00:07 RealBadAngel translating it to english, enable/disable switch doesnt disable shader itself but the effect done in particular shader which have to be run anyway 00:22 VanessaE in which case something will just need to be done to ensure the shaders always compile properly on as many hardware setups as possible 00:23 VanessaE is there a performance loss if e.g. bumpmap, waving leaves, waving grass are all enabled, but only, say, waving leaves are actually being used? 00:32 RealBadAngel if bumpmapping is disabled its code is simply not executed 00:32 RealBadAngel but remember when shaders are enabled each pixel light is calculated in shaders, not in the engine 00:33 RealBadAngel this is the main reason why any particular shader cannot be just simply disabled 00:34 RealBadAngel all the shaders, no matter the drawtype share same lighting code 00:45 VanessaE right 00:45 VanessaE so about getting them to compile... 00:45 VanessaE what is it that's broken on ShadowNinja's machine that isn't on mine, yours, exio's? 00:51 RealBadAngel i know whats wrong, his opengl is much more restrictive 00:51 RealBadAngel it demands a variable to be set 00:51 RealBadAngel if defined 00:52 RealBadAngel i will fix it right away 00:53 RealBadAngel my version pays no shit about that 00:54 RealBadAngel sometimes i think i should buy a dozen of PCs, with dozen of different GPUs and a few systems installed on each machine to debug standarized one language lol 00:56 RealBadAngel switching to linux, brb 01:03 VanessaE that's a long reboot time :P 01:04 Fury yeah 01:04 VanessaE RealBadAngel: that's like the lighting issue on mine - it allows math "overexposure" or something, so you had to clamp() it to make it match yours. 01:07 RealBadAngel ShadowNinja, ive pushed some updates, please do try now 01:07 RealBadAngel liquid shader shall work for you now 01:08 RealBadAngel ive also added versioning to each shader 01:12 VanessaE RealBadAngel: you need a better computer. 11 minutes is a fucking long reboot time ;) 01:16 RealBadAngel VanessaE, ive made fixes in shaders in those 11 minutes and pushed them :P 01:17 RealBadAngel also i started to revamp settings menu 01:17 VanessaE I know :) 01:18 RealBadAngel simply dissapearing of settings is not good imho 01:18 RealBadAngel i want them grayed out 01:18 VanessaE yes 01:19 RealBadAngel maybe leave setting desc, but grayed and no checkboxes? 01:21 RealBadAngel btw, any particular reason why this is still not merged? https://github.com/minetest/minetest/pull/385 01:23 VanessaE greyed out, fake grey checkboxes 01:23 VanessaE else it looks like a bug. 01:24 VanessaE "celeron55 commented 11 months ago -- Why?" 01:24 VanessaE that's why not :P 01:24 RealBadAngel answer for such question is usually "just because" 01:25 RealBadAngel i tried it and it feels good 01:26 RealBadAngel one of those changes that make world feel like moving 01:27 RealBadAngel same is for wavin leaves, small change but drastically change perception of the world 01:27 RealBadAngel sum of such small changes gives player feel that he plays real world 01:28 Fury well many people did like it and it was proposed with an option to turn off 01:30 RealBadAngel as almost everything here 01:31 Fury if a change is presented with the option to turn it off and every single comment, except with Calinous was " I agree, I want it" add it 01:31 RealBadAngel thus i can see no point to bake this pull any longer, 11 months is enough 01:32 RealBadAngel or we are planning this for 0.5? ;) 01:35 Fury no idea 01:36 RealBadAngel due to code freeze i wont be nagging to merge it now 01:36 RealBadAngel but just after release... 01:37 RealBadAngel ShadowNinja, here? 02:05 VanessaE you know, it occurs to me... if DanDuncombe can be convinced to get his fork of Realtest to a stable point, it would be a good candidate for including in official builds. 02:06 VanessaE (recalling previous discussions on the subject and the general consensus that more games need to be included with the engine) 02:07 RealBadAngel we already tried that 02:07 RealBadAngel and we ended with fight what shall be included in each one 02:08 VanessaE RealBadAngel: the previous attempt (survival/build/common) was a fiasco. This will be done differently, if at all. 02:08 VanessaE separate, self-contained games 02:08 VanessaE none of this cross-sharing stuff. 02:08 RealBadAngel to achieve that one condition shall be fulfilled in the first place 02:09 RealBadAngel different maintainers for each one 02:09 RealBadAngel with one folk ruling them all it will always fail 02:10 VanessaE yes, that's the plan 02:10 VanessaE my game has been proposed for inclusion, as has Pilztest 02:10 Fury did you think of a name? 02:11 VanessaE nope. 02:11 RealBadAngel PA shall too 02:11 VanessaE I still need help with that. 02:11 RealBadAngel otherwise folks can think its mushroom picking simulator 02:11 VanessaE haha 02:13 RealBadAngel and in the end they can think "damn, its so realistic, so hard, that i cannot find even a single mushroom" 02:14 Fury im playing in a minecraft server that has buildcraft and other similar mods. getting inspiration for minetest mods 02:14 RealBadAngel i wish some1 would bring there something similar to Thaumcraft 02:15 Fury this server has it and I still havent touched it 02:15 RealBadAngel its a good and well thought mod 02:15 RealBadAngel i liked to play with it 02:15 VanessaE whatever I name it, I don't want my game to have "mine", "test", or "craft" in the name. too many games like that already 02:16 RealBadAngel whatever the name 02:18 Fury well find something to call it fast so it can be included faster 02:18 RealBadAngel btw, have you heard about another voxel game using Irrlicht already? 02:19 RealBadAngel not an open source 02:20 RealBadAngel its pretty advanced 02:20 RealBadAngel http://www.buildaworld.net/ 02:22 Fury has better lightning 02:24 Fury wow the destruction of the building and the exploding volcano also look good 04:04 ShadowNinja hmmmm: Do you think we're ready to release? (As in tomorrow or the day after) 04:05 hmmmm if the Windows releases work then yes 05:14 VanessaE wait! 05:14 VanessaE don't release yet! 05:14 VanessaE apparently there's still that weird alpha issue celeron55 pointed out with leaves 05:14 VanessaE Sokomine reports this, with latest mt/mt_game: http://mg.viewskew.com/mgoblin_media/media_entries/457/screenshot_2203932552.png 05:15 Sokomine yes. and when i change perspective slightly (look a bit to the side/up or down/move a bit), it...flickers...kind of 05:17 VanessaE I don't see anything in mt_game that would cause this... 05:18 Sokomine see also: http://mg.viewskew.com/u/sokomine/m/screenshot-2204308095/ 05:19 Sokomine ...and http://mg.viewskew.com/u/sokomine/m/screenshot-2204312426/ (just looking a few degrees more to the left than in the first picture) 05:19 Sokomine with shaders disabled, the moretrees leaves look...odd. it flickers when switching perspective 05:19 VanessaE moretrees definitely don't do anything to cause this 05:19 VanessaE it has to be a bug in the allfaces_optional drawtype 05:22 Sokomine moretrees is innocent here. seems to be the tiny trees from bas080 that need an update here 05:24 VanessaE ok maybe not a bug in the engine thne. 05:24 VanessaE then* 05:25 VanessaE but c55 did find a similar problem with default trees 05:25 VanessaE so.. Is the z-ordering/sorting thing EVER gonna get fixed? 05:33 hmmmm uhhh 05:33 hmmmm 0.4.6? 05:33 VanessaE hmmmm: nope.avi 05:33 hmmmm I don't get it, why do you want to hold back 0.4.8 for a bug that's been around for a while and is clearly non-trivial to fix 05:33 VanessaE I still see z-ordering glitches here and there 05:33 hmmmm the point is 05:34 VanessaE I don't. The thing soko mentioned is not an engine bug apparently. 05:34 VanessaE it's a mod being stupid. 05:34 hmmmm you seem to think that minetest has to be perfect and completely bug free to release 0.4.8 05:34 Fury if it was non trivial it would have been fixed already 05:34 hmmmm but this same bug has been around before 0.4.7 even according to the screenshot 05:34 Fury but yeah release the kraken 05:34 VanessaE let me rephrase that - it's not a bug worth stopping the release for -- I thought it was the bug c55 complained about, which IS a showstopper. 05:34 VanessaE (or so he made it sound) 05:35 hmmmm what showstopper 05:35 VanessaE lemme find the ref. 05:35 hmmmm I am not aware of any transparency bugs aside from the alpha channel material bug which was added with the known caveat of bugginess 05:38 VanessaE http://irc.minetest.ru/minetest-dev/2013-11-10#i_3429508 05:38 hmmmm wtf 05:38 VanessaE yeah. 05:38 VanessaE look at the image in particular. 05:38 hmmmm and just as an aside I haven't seen anything like that 05:38 hmmmm so is it consistent for him? 05:39 VanessaE I've seen similar effects but only with clouds vs. water 05:39 VanessaE idk, that's the first and last I've heard from him about that 05:39 hmmmm and again 05:39 hmmmm it seems to be that one mod 05:39 hmmmm the "Eden" game 05:40 VanessaE oh? 05:40 hmmmm yes, read the context of that quote 05:40 Sokomine er, no, my screenshots where taken from latest git of everything relevant involved 05:40 Sokomine i have no clue why my mt says it's 0.4.6 05:40 VanessaE I guess I have to yell at Mossmanikin. Eden is his game isn't it? 05:41 hmmmm it's Neuromancer 05:41 Sokomine the problem i encoundered seems to be mod-specific. some textures have it 05:41 hmmmm I don't really understand why he's using alpha levels on leaves when he should be using allfaces_optional 05:41 Sokomine activating shaders removes the glitch, but that's not a solution for all and i'm not even sure i like shaders 05:42 hmmmm half of these supposed bugs come from people using modding features the wrong way 05:42 hmmmm now I regret adding use_texture_alpha. I really thought people could be mature about it and stick to only things like colored glass but I guess that's asking for too much 05:43 Sokomine no. just explain what's wrong. perhaps that particular modder didn't get the why 05:45 VanessaE hmmmm: the problem is some of this stuff has only ever been explained here or on some core dev resources -- but not to the modding community as a whole. 05:46 Sokomine it's bas080' tree mod that has the leaves problem. if you know what's wrong with those textures/node definitions it might be helpful if you could tell us. if you want to, you can post it in the relevant thread and explain. i'll have to write something regarding that mod later anyway 05:46 Sokomine *nod* quite a large part of what ought to be known is not. repeating an explanation can help 05:47 bas080 hmmmm, i'm here to push the envelope 05:48 bas080 I'll make sure to change it so it works for the majority. 05:48 hmmmm alrgiht 05:48 hmmmm it's really simple: 05:48 VanessaE hmmmm: that said, regarding long-standing hard-to-fix bugs.... so, are we grandfathering-in bugs now? I don't mean for the 0.4.8 milestone, I mean...just ever? 05:48 hmmmm see where you have "use_texture_alpha = true"? 05:48 hmmmm remove that 05:48 bas080 hmmmm, :P 05:49 hmmmm VanessaE, if someone can figure them out, then that's great, if not and they're not critical, then it's NO reason to hold up 0.4.8 05:49 hmmmm we've been waiting 5 months for what was supposed to be a 1.5 month release 05:49 VanessaE hmmmm: I agree - which is why I just said it wasn't for 0.4.8 but just, ever. 05:49 hmmmm and the reason was 90% because I went onward with my life and got a job and moved and shit and couldn't be around to nag you people 05:50 VanessaE heh 05:50 hmmmm you would never release 0.4.8 if I weren't here nagging 05:50 VanessaE don't nag me, I'm not a core dev so no one here listens to what I have to say anyway :) 05:50 Sokomine :-) hope it gets released soon. people who prefer the version labled stable ought to get the advantages recently added 05:52 Fury just release it already and stick to the deadline for xxx.9 next time 05:53 Fury :D 05:54 hmmmm we're waiting for thexyz and sfan5 to compile windows versions 06:34 VanessaE hmmmm: it just occurred to me...if the delay is now how long it takes before new builds are made...shouldn't the repo be tagged/marked for 0.4.8? 06:48 hmmmm no, I'm waiting to verify that builds can be made so that it can be tagged for 0.4.8, and they're recompiled as such then 06:48 VanessaE oh ok 06:48 hmmmm if they don't build then that's not really 0.4.8 now is it 06:48 VanessaE true 06:48 hmmmm i'd have to rewrite history to undo that 06:53 VanessaE I'd give it a shot but the newest windows I have on hand is win2k :) 07:06 thexyz there's the dev wiko 07:06 thexyz wiki* 07:06 thexyz you complain about people using featured the wrong way 07:06 thexyz yet neither right nor wrong way is documented 07:25 sfan5 hmmmm: ? 07:25 sfan5 hi everyone btw 07:26 VanessaE hey 07:28 us-0gb Hello! 08:32 thexyz time for some daily dose of WAT 08:33 thexyz https://github.com/minetest/minetest/blob/master/src/map.cpp#L1817 warning C4804: '>' : unsafe use of type 'bool' in operation 08:33 thexyz https://github.com/minetest/minetest/blob/master/src/main.cpp#L1256 warning C4129: '%' : unrecognized character escape sequence 08:35 thexyz cool 08:35 thexyz findcurl overrides my manually specified curl dll path 08:54 thexyz here's the build http://a.pomf.se/gdc1r.zip 10:20 celeron55 thexyz: the first one is clearly a bug; NodeNeighbor::l is a boolean 10:20 celeron55 and the second one... what 10:21 celeron55 why is the percent sign escaped 10:25 celeron55 it comes from the original leveldb commit 12:49 PilzAdam thexyz, can you merge the translations in? 14:26 nore any thoughts on this? https://github.com/minetest/minetest/pull/1017 14:31 sfan5 opinions please! https://github.com/minetest/minetest/pull/1018 14:31 sfan5 nore: document the packet format in clientserver.h 14:31 nore sfan5, I'm ok with it 14:31 nore oh, yes, forgotten that 14:32 sfan5 I guess I'll merge #1018 then 14:33 sfan5 I wish Github would include a button for merging without merge commit 14:34 PilzAdam sfan5, ummm... we have a feature freeze 14:34 sfan5 that is not a feature 14:34 sfan5 more like a precaution 14:34 sfan5 many people just enable everything and then wonder why it does not work 14:35 PilzAdam committing something hours before a release that isnt a high priority bugfix isnt good 14:36 nore sfan5, done 14:36 sfan5 :-/ 14:36 nore PilzAdam, is 0.4.8 coming out? (at last...) 14:36 sfan5 yes 14:36 nore PilzAdam, what about the swap_node thing (for 0.4.9) 14:37 sfan5 #1018 is just a tiny bugfix/precaution, I'm going to push.. 14:38 PilzAdam sfan5, it adds translations, so we have to wait for all languages to translate it first 14:38 PilzAdam that will take at least another day 14:38 PilzAdam do you really want to block 0.4.8 because of this? 14:38 nore sfan5, I'm wondering... hasn't enable_shaders be changed to use a bool value? 14:38 sfan5 not every language has everything translated 14:38 sfan5 nore: yes 14:38 sfan5 but 0 is false 14:39 nore it should set "false" then... 14:39 PilzAdam sfan5, http://translate.minetest.ru/projects/minetest/core/ there are quite a few with 100% 14:39 PilzAdam sfan5, dafuq! 14:39 PilzAdam can you stop pushing things if 2 core devs said "no"? 14:39 sfan5 2 core devs? who? 14:40 PilzAdam nore and me 14:40 sfan5 sfan5, I'm ok with it 14:40 nore (I'm only asking to set to false and not 0) 14:40 sfan5 that means no? 14:40 nore sfan5: I'm ok if you change the 0 to false 14:41 sfan5 anyway, we don't need to delay 0.4.8 because onyl 99,9% is translated 14:41 sfan5 nore: thats kinda pointless, but ok 14:41 nore and about translations: I guess asking in #minetest channel should be ok to have half of them in less than 1 hour 14:42 sfan5 ok, fixed 14:42 PilzAdam nore, no, I tried that, it takes at least a day to get ~4 translations 14:42 PilzAdam sfan5, we dont add () arround conditions in Lua 14:42 nore does it? we can have 2 translations right now (French and German) 14:42 PilzAdam we first need to wait for thexyz to update weblate 14:43 sfan5 PilzAdam: are you expecting me to change that too? 14:43 PilzAdam sfan5, I would have expected you to fix it _before_ pushing it 14:43 nore PilzAdam, is it ok if I add #1017 to 0.4.9 milestone? 14:44 sfan5 you told me after I rebased & pushed nore's fix 14:44 sfan5 s/fix/suggestion/ 14:44 sfan5 s/suggestion/fix/ 14:44 PilzAdam nore, I have no idea what the 0.4.9 milestone is meant for 14:45 PilzAdam sfan5, you should have waited longer before pushing 14:45 nore PilzAdam, " 14:45 nore Features to be added to 0.4.9. they may not be ready for merge yet. Prior merging pulls need to be checked for completeness." 14:45 nore there: https://github.com/minetest/minetest/issues/milestones 14:46 PilzAdam nore, if you add all other pull requests and issues that fit this description too, then ok 14:46 sfan5 PilzAdam: can you update the .po files and commit? 14:46 PilzAdam sfan5, no, I dont know if that would mess up weblate 14:47 sfan5 no, weblate is updated manually... oh.. ok 14:47 sfan5 we can only do that if thexyz pushes the current translations from weblate 14:47 sfan5 (if there are any) 14:49 sfan5 PilzAdam: did you take a look yet? https://github.com/minetest/minetest/pull/958#issuecomment-26757928 14:49 sfan5 I'd relly like to get a player list into 0.4.9 and formspec customizeable would be nice too 14:49 sfan5 really* 14:51 PilzAdam for now Id like to focus on getting 0.4.8 out 14:51 sfan5 ok 14:52 nore is the milestone ok? https://github.com/minetest/minetest/issues?milestone=2&state=open 14:52 nore (anything that I have forgotten or that shouldn't be there?) 14:52 sfan5 that looks ok 15:04 nore sfan5, what doesn't work with that? https://github.com/sfan5/minetest/commits/tab_playerlist2 15:04 sfan5 segfault 15:05 nore when? (server start, when pressing tab, ...) 15:06 sfan5 pressing tab 15:06 kahrl Can I fix the invalid escape sequence in main.cpp:1256? Seems trivial enough not to interfere with the code freeze 15:07 nore sfan5, you should perhaps send the string to a client when connecting to the server... 15:07 nore kahrl, looking at it 15:08 nore yes, you should fix it... 15:08 kahrl ok, I'll push it 15:08 nore it seems indeed trivial 15:09 sfan5 nore: it is designed to allow a custom player list if wanted, if no string is get by the client it will display the default one 15:09 sfan5 just add a test mod that does set_playerlist for every player 15:10 nore sfan5, yes, but the string should nevertheless be sent by the server (i.e. the server starts, changes the string, and then I join) 15:10 nore or is it per_player? 15:10 sfan5 yes, per player 15:11 nore and do you know where the sergfault lies? 15:11 sfan5 to allow '''This way mods can group the list to teams or something similiar.''' 15:11 sfan5 yes in Irrlicht IIRC 15:12 nore couldn't it be some formspec-related thing then? (I guess perhaps a formspec being empty, or something like that) 15:13 sfan5 let me reproduce it and give you a backtrace 15:24 sfan5 now it does not work at all 15:26 EvergreenTree Ahem, sorry for repeating myself, but no one noticed when I reported this: 15:26 EvergreenTree I have noticed a problem with the inventory image for the jungle sapling: http://i.imgur.com/MhJjAI1.png 15:26 EvergreenTree Since it is pretty much the same code as the regular one, it shouldn't be any different 15:26 EvergreenTree I think VanessaE had a problem with this, I think it has something to do with the plantlike drawtype 15:26 EvergreenTree It is especially strange, as the regular sapling looks normal, but the jungle one doesn't. They are practically the same code, so I have no idea 15:29 nore EvergreenTree, is the image the same? I thought it had always been like that... 15:30 EvergreenTree IT has, which is a problem 15:30 EvergreenTree Considering the default sapling looks normal 15:30 EvergreenTree In comparison 15:30 EvergreenTree VanessaE had a problem with this in moretrees, where the leaves would be squashed when the drawtype is set to plantlike 15:31 * EvergreenTree pokes VanessaE 15:31 EvergreenTree I didn't fix it, so I'm not sure how 15:36 adama_ hello every one 15:47 nore hey, something that should probably be done before 0.4.8: buckets should check area protection 15:47 sfan5 nore: when i press tab the mouse unfocuses and pressing esc after that segfaults 15:48 nore strange... 15:48 nore sfan5, what about ^ 15:48 sfan5 that counts as feature and we have feature freeze 15:49 nore a feature? I'd count it a bug... 15:51 sfan5 backtrace: http://pastie.org/8503458 15:51 sfan5 ^ nore 15:51 nore no idea then... 16:01 nore https://github.com/minetest/minetest_game/pull/221 16:08 adama_ VanessaE is your plantlife mod compatable with dev. version of MT? 16:09 pitriss adama_: of course it is.. 16:10 adama_ ok 16:10 pitriss adama_: but for mod related question you should use non dev channel 16:10 nore adama_, don't ask modding-related things here, this channel is for engine only 16:10 adama_ lol 16:10 adama_ sorry 16:10 nore np 16:10 adama_ thank you 16:19 nore so, releasing 0.4.8 soon? 16:38 PilzAdam thexyz, https://github.com/PilzAdam/minetest/commit/fb150e05043022ff2bb96ea892a31c632f130aca 16:38 PilzAdam can I push that and will you update weblate? 16:39 thexyz well if it doesn't break anything then do that 16:40 PilzAdam done 16:45 thexyz updated weblate as well 16:46 PilzAdam and german translation done 16:46 thexyz anyone actually tested my windows build? 16:46 PilzAdam yes, works in wine 16:47 thexyz so after the most common use case is covered anyone tested it on windows? 16:47 nore PilzAdam, I can do translation in French if you need it 17:01 Calinou me too 17:03 Calinou done :> 17:15 sfan5 I'd like to have #1019 and #1020 fixed before we release 0.4.8 17:16 PilzAdam sfan5, "rename button"? 17:17 sfan5 you can see the button on this screenshot: https://f.cloud.github.com/assets/1042418/1606914/c2186ad8-5462-11e3-834b-bd6b6bee69d8.png 17:18 nore sfan5, is that modmgr? 17:18 PilzAdam there is also the issue with the wrong world being selected in a long world list 17:19 sfan5 nore: yes 17:19 PilzAdam sfan5, I reported #1020 to sapier a while ago 17:41 nore sfan5, could you try replacing the modmgr.modlist by modmgr.global_mods for #1019? 17:41 nore (there are a few ones to replace for the rename modpack thing) 17:42 sfan5 one sec, I'm just testing another fix 17:43 nore there are exactly 2 of those 17:43 nore it looks like modmgr.modlist is for world use only, not for global use 17:44 nore ok, now I am sure it has to be that 17:45 sfan5 2 of them? 17:45 nore yes, the one that crahsed and another one 17:45 nore in modmgr.handle_rename_modpack_buttons(fields) 17:46 nore sry, I don't have the correct lines numbers here, I'm not on the most recent version 17:46 nore ah, I can test that as soon as compiling is finished... 17:46 sfan5 wow, a Debug build is really slow 17:47 nore why do you use it? 17:47 sfan5 works perfectly 17:47 sfan5 to test the tab_playerlist2 issue 17:48 sfan5 https://github.com/minetest/minetest/pull/1021 17:49 PilzAdam oh, the bug that the wrong world is selected only happens when I use keys to navigate to it 17:49 nore sfan5, it doesn't for me 17:50 PilzAdam which reminds me that kahrl's formspec_table has to go in sometime soon 17:50 nore I don't get crash, but the modpack is not renamed 17:50 nore PilzAdam, I added it to 0.4.9 milestone 17:51 nore sorry, it works 17:51 nore but the modname is not changed in the menu 17:51 nore (it is not updated) 17:53 PilzAdam sfan5, commit messages shouldnt be "Fix #xxx" 17:56 nore sfan5, at least add that line 679: modmgr.refresh_globals() 17:56 nore I'm working out how to fix mod not being correctly selected after that 18:01 sfan5 nore: pull updated 18:01 nore wait a bit though 18:03 nore ok, add that after it: 18:03 nore modmgr.selected_mod = filterlist.get_current_index(modmgr.global_mods, 18:03 nore filterlist.raw_index_by_uid(modmgr.global_mods, fields["te_modpack_name"])) 18:03 nore that will make the modpack you just renamed be selected again 18:03 nore after that, I'm ok with the pull... 18:04 sfan5 updated... merging 18:05 nore sfan5, what's the other bug? 18:05 sfan5 #1020 18:08 nore ok, I may have the answer 18:08 nore trying to fix it 18:22 nore there is an important bug that crashes it: 18:23 nore I added print(5.2) line 328 of mainmenu.lua 18:23 nore here it what I get: 18:23 nore 5.2 18:23 nore Loaded texture: /home/nathanael/Minetest/textures/base/pack/no_screenshot.png 18:23 nore Loaded texture: /home/nathanael/Minetest/textures/base/pack/menu_header.png 18:23 nore 5.2 18:23 nore 5,2 18:24 nore ^ the last thing is what crashes everything: Lua numbers got a "," instead of a ".", and the formspec size element gets corrupted 18:24 nore sfan5, ^ 18:24 nore I have absolutely no idea about how this could happen, though 18:25 sfan5 wat 18:25 nore you do think is it impossible too, don't you? 18:25 sfan5 yes 18:26 nore yet it is (maybe locale? how are decimal numbers represented in German? are they with a comma too?) 18:26 nore (and do you get "invalid size element" too?) 18:27 sfan5 YES 18:27 sfan5 s/YES/yes/ 18:27 nore yes for the decimal comma? 18:28 sfan5 yes 18:28 sfan5 but I don 18:29 sfan5 't see how locale could affect internal strings 18:29 nore the strange thing is that gettext isn't even enabled on my build... 18:30 nore sfan5, is doesn't affect internal strings, it affects how numbers are displayed 18:30 sfan5 hmm 18:30 nore but I don't see why a call to show_file_open_dialog would cahnge something 18:33 kahrl nore: does the bug happen in 0f9440fa61f? 18:33 nore kahrl, do you have the bug? 18:33 kahrl no 18:34 nore how do I go back to that? 18:34 kahrl git checkout 0f9440fa61f and rebuild 18:36 nore kahrl, "error: pathspec '0f9440fa61f' did not match any file(s) known to git." 18:36 kahrl ah, I managed to get the bug now with LANG=de_DE ../bin/minetest 18:37 nore kahrl, what is strange is that gettext isn't even enabled in my build... 18:38 kahrl well even without gettext there is some code that sets the numeric locale 18:38 kahrl some of that must be buggy 18:39 kahrl I got the bug in a build with gettext, so that build option has no effect 18:39 nore but why do we even set numeric locale? 18:40 kahrl it should always be set to "C" 18:40 kahrl but I think there is some code that sets LC_ALL, maybe it forgets to set LC_NUMERIC 18:41 nore src/guiFileSelectMenu.cpp: m_previous_locale = setlocale(LC_ALL,0); 18:41 kahrl weird, I can checkout 0f9440fa61f no problem 18:41 nore src/guiFileSelectMenu.cpp: setlocale(LC_ALL,m_previous_locale.c_str()); 18:41 nore kahrl, must be that.. ^ 18:41 kahrl oh indeed 18:41 kahrl there we go =) 18:44 kahrl wait, those line are there in 0f9440fa61f but got removed in 22a59b3912f 18:44 kahrl lines* 18:44 nore yes, I saw that... 18:45 nore grep -R "LC_ALL" only outputs gettext.cpp, lua/src/loslib.c and intlGUIEditBox.cpp 18:46 kahrl ugh, there's a setlocale(LC_ALL, "") in CGUIFileOpenDialog.cpp 18:47 kahrl enabled on all platforms except windows CE 18:48 nore where is it? 18:48 nore (I can't find the file) 18:48 kahrl irrlicht-1.8/source/Irrlicht 18:49 nore ah, in Irrlicht... what do we do then? 18:49 kahrl work around it as usual ;) 18:50 nore that is? 18:50 kahrl I think the lines that you posted above were such a workaround, do you know why sapier removed them? 18:50 kahrl of course the workaround failed to reset LC_NUMERIC but that could be fixed 18:50 nore no idea... but LC_NUMERIC should be set too 18:51 nore do you know in what commit he removed them? 18:51 kahrl 22a59b3912f 18:52 nore I'd say he thought they were not needed anymore... 18:52 kahrl maybe 18:52 nore should I label the bug as blocker? 18:53 kahrl dunno... is it really that big of a deal? 18:53 nore dunno... could you try to put those lines here again, with the LC_NUMERIC one too? 18:53 kahrl especially since you can just click on another tab to fix the menu 18:54 kahrl I'll try 18:54 nore no, you can't 18:54 kahrl oh, I"m on the old commit still 18:55 kahrl it does work there 18:56 nore the proper thing to do would be to save all locale things 18:56 nore and restore them at the end, not only the numeric one 18:58 nore grrr, why doesn't Irrlicht accept utf8? 18:59 kahrl the workaround plus setlocale(LC_NUMERIC,"C") don't fix it 19:02 nore kahrl, can I merge that? https://github.com/Novatux/minetest/commit/e14c9759e276b2fc3e29004732bbbb81d87eaf6d 19:02 OldCoder I have time today to rebuild servers and clients. Is upstream git in a state where people recommend I use it for live servers? 19:03 kahrl nore: yes 19:03 kahrl other people are written without umlauts too (Jürgen, Neuschäfer) 19:04 kahrl OldCoder: wait for 0.4.8 maybe 19:04 OldCoder kahrl, Very well. It is intended for December, correct? 19:04 kahrl I hope it doesn't take that long... 19:04 OldCoder Have fun :-) 19:05 OldCoder If you need testing, let me know 19:05 kahrl sure 19:05 OldCoder kahrl, has anybody else reported the lockup on item definitions issue? I do not have many people besides myself to test for that. 19:05 OldCoder Sapier believes it is shaders 19:06 sfan5 kahrl: hmmmm wants to have 0.4.8 today or tomorrow 19:06 kahrl I haven't seen anything about that 19:06 nore kahrl, yes, I didn't noticed it before... I hate Irrlicht for that 19:06 kahrl sfan5: hmmmm wanted it a week ago ;) 19:06 OldCoder Ah. 0.4.8 may be relatively soon. 19:06 sfan5 anyay, it would be nice if we release 0.4.8 today 19:06 OldCoder ! 19:07 nore kahrl, have you been able to fix that bug, or not? 19:07 sfan5 anyway* 19:07 kahrl nore: the numeric locale bug? no... 19:08 nore and did you try to set the locale right before returning to Lua? 19:08 nore I guess it should work, now you set it when exiting the menu 19:08 nore but a Lua call is executed before, I guess 19:09 nore do you see what I mean? 19:09 kahrl oh, yeah 19:09 kahrl I was under the impression show_file_open_dialog blocked until the dialog returned 19:10 nore wait, even better: where is the setlocale in the code? 19:11 nore the best would be to undo it right after that 19:11 kahrl well it's inside irrlicht 19:11 nore yes, but in what call? 19:11 kahrl happens during the constructor as well as several events 19:11 kahrl seriously, wtf 19:12 nore if I were you, I would make a macro to reset correctly the locale 19:12 nore and add it everywhere... ;) 19:12 kahrl Raymond Chen would say "Don't use global state to manage a local problem" 19:12 nore seriously, we need to report that to the Irrlicht guys... 19:12 kahrl setting the locale is slow... I really don't want to spam setlocale everywhere 19:13 nore slow how? 10ms? 19:13 nore more? 19:13 kahrl no idea 19:13 nore cause if it is less than 10ms, it is no problem to set it everywhere... it is a menu, that is not opened often 19:16 kahrl I don't know if doing it in the menu is enough 19:16 nore I would do it in destructor + each time Lua is called 19:17 nore I guess it is enough since in the old commit, refreshing the formspec fixed the bugs, as you said 19:18 kahrl well refreshing it still works for me in the current version 19:18 kahrl each time Lua is called -> that might be quite often? 19:18 nore no, not that much... only each time you send an event 19:19 nore and that is not *that* frequent 19:19 nore moreover, I guess that a Lua call is longer that just setting the locale... 19:21 kahrl it works, but I hate that it has to be done because of a rarely used menu 19:21 nore yes, I hate it too... but we have no other choice, I guess 19:21 nore we need to say something to the Irrlicht maintainers about that 19:23 kahrl apparently others have run into it as well 19:23 kahrl http://www.irrlicht3d.org/wiki/index.php?n=Main.ConvertingStringsToVectorsIntegersAndFloats 19:26 nore kahrl, I found that in a forum post: It's only my low-level file-access functions that need to force it temporarily back to English settings to read and write the files correctly. 3 x setlocale() is not too much overhead each time a file is written or read. The performance of the file IO will outweigh the small overhead of setlocale(). 19:26 nore so I guess setlocale is not very time-hungry 19:27 kahrl but that's compared to file IO 19:28 nore yes, I know... but you don't click 10 times a second in the window either... and Lua calls are not very efficient either 19:31 nore but anyway, you said locale was frequently set by Irrlicht 19:31 nore that means it is not too long to set it, else we would find it slow... 19:34 nore kahrl, so, what do we do? 19:35 kahrl does sapier usually come by at this time? 19:35 kahrl he's been working with all this locale stuff in minetest so I'd like to know what he says 19:36 nore I don't know... I reckon he usually conencts earlier 19:38 hmmmm hhh 19:38 hmmmm forget about locale problems for now, besides, getting text to render properly under irrlicht is a farce 19:39 hmmmm I think we should release really soon, whenever we get the final word on MSVC and mingw compatibility 19:39 nore hmmmm, the problem is that Irrlicht changes the numeric locale in the file select menu, which causes the formspec to crash 19:39 hmmmm that's great 19:40 nore after a call, since the size elements gets from size[12,5.2] to size[12,5,2] 19:40 hmmmm yeah, sapier was warning about this 19:40 hmmmm what he wanted to do was to wrap every single atof() call with a setlocale 19:40 PilzAdam hmmmm, the MSVC build works 19:41 hmmmm I think that for now, the irrlicht file select dialogs should NOT be used 19:41 hmmmm I don't get who's using it or why 19:41 nore so, code it again? 19:41 nore it is used in mod manager 19:41 hmmmm oh that's great 19:41 hmmmm why does irrlicht need to be such a broken turd 19:41 nore perhaps copy the code from Irrlicht then, without the locale setting thing could work? 19:42 nore kahrl, dunno... 19:42 hmmmm that would be a way if the irrlicht dialog uses all external interfaces 19:45 nore source here: https://www.assembla.com/code/myirrlicht-gitsvn/git/nodes/33c8730aa3f0bf2786ceb5b2210af7ffaa033f39/source/Irrlicht/CGUIFileOpenDialog.cpp 19:45 nore do you think it can work without being in Irrlicht? 19:45 PilzAdam does it work with Irrlicht 1.7 and 1.8? 19:46 nore eh, no idea 19:47 nore anyway, last commit on that file looks like 3 years ago... 19:47 nore but last commit at all is 3 years ago too 19:50 kahrl use the sourceforge svn repository 19:51 hmmmm alright, so it changes the locale at the end of the ctor and OnEvent 19:52 kahrl nore, that version you linked doesn't even restore the working directory ;) 19:52 hmmmm is it possible we can fix this by somehow hooking OnEvent of irrlicht's window for the dialog 19:52 nore yes, I guess that... It looks like it is very old 19:52 hmmmm we set the locale back after the object is created and inside the hook, provided it's run after the parent OnEvent 19:52 nore the Irrlicht locale is only changed when creating the list of files 19:52 hmmmm see I don't know enough about irrlicht's gui to know if we can do that.. can we? 19:53 kahrl not sure if the patch that restored the working directory is in irrlicht 1.7 19:53 nore so changing it back when we get input should work 19:53 nore kahrl, then copy the 1.8 version, and test if it works with 1.7 19:53 nore if it does, we can keep the 1.8 one... 19:54 kahrl CGUIFileOpenDialog::OnEvent returns true after calling fillListBox, so it eats the event 19:54 hmmmm nore, setting it on input is bad because what happens to the things that use locale while the dialog is open 19:54 kahrl maintaining our own version or resetting the locale on every lua call seem to be the only options 19:54 hmmmm I agree 19:55 hmmmm just wanted to see if there was a way to avoid that 19:55 nore I'd prefer the first one if it works, else, we don't have choice... 19:55 kahrl but our own file dialog is kind of heavy for a code freeze 19:55 hmmmm by the way, what other things happen if we define _IRR_WINDOWS_CE_PLATFORM_? 19:56 hmmmm yeah I agree, this is a pretty bad bug though 19:56 hmmmm modmanager wasn't in 0.4.7 when that was released, so if it comes out now with such a major issue that wouldn't be great either 19:57 nore kahrl, on the other hand, it is not ours, it is only mocing it from Irrlicht ot here... 19:57 kahrl that's true 19:58 nore why does it even change the locale? 20:00 kahrl I can't tell, sf.net is giving me HTTP 500 :P 20:05 kahrl http://sourceforge.net/p/irrlicht/patches/86/ no comment on setlocale 20:15 nore and have you tried to remove it to see what happens? 20:18 nore kahrl, ^ 20:18 kahrl no 20:19 kahrl we use the default locale to LC_CTYPE anyway so mbstowcs wouldn't change 20:19 kahrl set* 20:19 kahrl err, use* for* 20:20 kahrl I don't really see the point of setlocale there, it probably was added because the program of the one who made the patch didn't set the locale properly 20:26 nore ok... I can't get it to compile no, though 20:26 nore virtual const wchar_t* getFileName() const _IRR_OVERRIDE_; 20:26 nore +w 20:27 nore it crashes in the h file... 20:27 nore what is that _IRR_OVERRIDE_ ? 20:28 kahrl I got an idea from #irrlicht 20:28 kahrl what if we simply call putenv("LC_NUMERIC","C") 20:28 VanessaE ...and you guys wanted to push this out after only a week of feature freeze? :) 20:28 kahrl err putenv("LC_NUMERIC=C") or the setenv equivalent 20:28 nore kahrl, what's putenv? 20:29 kahrl sets an environment variable 20:29 kahrl setlocale(LC_ALL,"") takes the locale from the environment 20:29 nore quite clever... 20:29 nore I want to see if this works 20:33 nore kahrl, so? 20:36 kahrl https://gist.github.com/kahrl/7619600 20:36 kahrl works on gentoo linux with: LANG=de_DE ../bin/minetest 20:38 kahrl I don't know if windows's setlocale cares about the environment at all 20:39 nore kahrl, I hope they do... (you should check that with wine...) 20:40 kahrl still don't have that installed 20:40 PilzAdam VanessaE, even if we release at christmas, we will always find something that can be fixed 20:40 nore ok, works well on Debian 64bit, with FR as system default locale 20:41 nore I guess we only need someone to check the windows versions... 20:41 nore (sfan5 perhaps?) 20:41 sfan5 hm?7 20:41 sfan5 -7 20:42 nore could you test that fix for windows versions? https://gist.github.com/kahrl/7619600 20:42 VanessaE PilzAdam: perhaps :) 20:42 nore (since you make windows builds...) 20:42 VanessaE (insert the old axiom about every program being reducible to one line that doesn't work) 20:43 ShadowNinja nore: You add whitespace to the end of a line in your minetest_game buckets pull. 20:44 nore where? 20:44 sfan5 nore: I'll test 20:46 nore ShadowNinja, is it ok now? 20:48 ShadowNinja nore: Yes, alhough I would like it if you could reduce the duplication. 20:53 sfan5 nore: does not worj#k 20:53 sfan5 work* 20:53 nore :( 20:53 nore kahrl, ^ 20:54 nore ShadowNinja, like that? 20:55 ShadowNinja nore: Yes. 20:55 nore ShadowNinja, should it be merged before/after 0.4.8? 21:10 nore kahrl, so, what do we do since it doesn't work on windows? 21:38 proller fix bool > 0 - https://github.com/proller/minetest/commit/7e575cfc6a97f7e39df616a7a5a39447062468d0 21:40 nore what would you think about adding the bit32 Lua library as default Lua available libraries? (it is in standard Lua 5.2, but not 5.1) 21:43 sfan5 that is a good idea 22:00 PilzAdam ShadowNinja, the long dig time in creative mode is ridiculous, do you really want to release it? 22:01 ShadowNinja PilzAdam: Yes, and it was fine last release. 22:02 PilzAdam someone should update the changelog in the dev wiki 22:05 ShadowNinja PilzAdam: So is it mitori or mito551? 22:05 PilzAdam ShadowNinja, dunno, ask him