Time Nick Message 00:30 paramat hi celeron55, i also want hmmmm back as a core dev, at any cost, please invite him. nerzhul was promoted too quickly, is rather arrogant and evasive and disrespects hmmmm's experience. nerzhul is talented but needs to be more humble, possibly demoted to a contributor, not a core dev. anything that forces hmmmm out is a disaster for Minetest 00:37 paramat sorry to disturb you so late, it's probably too late to discuss, just wanted to express my opinion =) 00:38 est31 that network patch was the reason he left? 00:38 est31 wow. 00:42 paramat not entirely 00:45 paramat (obviously i can't speak for him) 00:47 hmmmm there are a lot of reasons 00:50 hmmmm i mean what are you supposed to do when some totally unknown guy just drops in one day, is made a committer (obviously he has the backing of celeron), says phrases like "we need to move fast, it's okay if things get broken, this is the agile way", speaks on behalf of all the developers obnoxiously using the pronoun "we" to voice his own opinions and plans he made up completely on his own with no input from others 00:50 hmmmm and then when you try to give input he ignores you 00:51 hmmmm it's like it's some sort of hostile takeover 00:52 paramat it was really good to see you tackle things earlier, i was not impressed with his attitude 00:52 paramat you had amazing plans for sorting MT out, i would love to see you realise them here, instead of forking 00:52 hmmmm for a couple of days i went along with it, like w/e 00:53 hmmmm then i got tired of that shit and gems like "you need to work on this on sunday" - like what, is minetest a real job, and he's our manager? 00:54 hmmmm i was just so alienated at that point I just said I'm out 00:54 hmmmm about a week after I 'quit' and everything I realized it was a mistake because there were at least 3 others sharing my same sentiment 00:57 est31 nrzkt's native language isn't english. I dont think he really means it when he says things like that. not saying he can't improve. 00:58 hmmmm calinou's native language isn't english and there's no issue 00:58 est31 e.g. he sais "must", where "should" is appropriate 00:58 est31 and I think he actually means should 01:02 paramat it's not just language issues, nerzhul's attitude is clearly arrogant and disrespectful, i can see that now 01:07 paramat i would make hmmmmm project leader and nerzhul a contributor for the short term, under authority of someone we trust, if nerz has the correct humble attitude s/he will accept that 01:08 est31 that sounds reasonable, I support it. 01:08 est31 what do you mean with contributor? 01:08 paramat not a core dev 01:09 est31 ah ok 01:09 est31 I think nerzhul deserves a second chance. 01:11 paramat i worked full time for 2.5 years on mapgen, then unexpectedly hmmmmm and i swapped places, but i still consider him a sort of authority, that's the correct attitude 01:11 paramat yes i agree 01:13 est31 have you changed nicks recently paramat? 01:13 est31 excuse me, I'm very new to this. 01:13 luizrpgluiz paramat: contribute to the game is very important 01:14 luizrpgluiz :) 01:15 paramat of course luiz 01:15 paramat est, um no, just acquired my yellow spot =D 01:16 paramat and registered with nickserv 01:16 luizrpgluiz Paramat is master in mapgen 01:16 est31 ah I see... 01:17 paramat lol 01:17 luizrpgluiz :D 01:17 paramat i guess someone new and promoted too quickly hasn't had time to build up respect for those who deserve it 01:18 paramat luiz hmmmmm is just as good and probably better 01:18 est31 In theory, you should respect anybody, regardless of their titles. 01:19 est31 but yes esp those who have deserved it 01:20 luizrpgluiz yes, but with each new version, changes rather mapgen 01:20 paramat heh c55 is probably asleep. i'll let him know my opinion though 01:20 paramat i'm not going to watch hmmmmm go without fighting to keep him! 01:25 luizrpgluiz paramat: someday you can create a mapgen only commenting each line? 01:26 luizrpgluiz :D 01:28 paramat luiz best discuss this in minetest channel 03:08 est31 ShadowNinja, #2482 03:08 ShadowBot https://github.com/minetest/minetest/issues/2482 -- Fix bug that disabled game-based minetest.conf default setting by est31 03:11 ShadowNinja est31: Commented. 03:11 est31 seen 03:11 * est31 checks how to solve it 03:36 est31 ShadowNinja, check #2482 again 03:36 ShadowBot https://github.com/minetest/minetest/issues/2482 -- Fix bug that disabled game-based minetest.conf default setting by est31 03:40 ShadowNinja est31: Much better. :-) 03:47 est31 ShadowNinja, have you seen my comment? 03:47 est31 answer rather 03:48 est31 I think loading config isn't part of world initialisation 03:48 est31 as it affects global variables 03:49 ShadowNinja est31: Well, I suppose it isn't. 03:49 ShadowNinja Seems like the logging should be at the top of the function though. 03:50 est31 because? 03:54 est31 hmm, found next bug... settings shouldn't probably be global. when you load subgame A with game config a, and then load subgame B, it loads using config a as default. 03:54 est31 not a large or important one 03:55 est31 therefore I won't fix it :D 04:14 est31 ShadowNinja, see my comment on your security-2 branch 04:14 est31 (not related to the discussion bout my PR) 04:17 ShadowNinja est31: Fix that by re-running set_default_settings before loading the overrides. 04:19 est31 ShadowNinja, that isnt enough, they need to be emptied completely 04:20 est31 imagine custom settings 04:22 ShadowNinja est31: Make set_default... clear all defaults first. 04:23 est31 you mean clear every setting= 04:23 est31 ?* 04:24 ShadowNinja est31: No, just the defaults. 04:25 ShadowNinja est31: Defaults and explicit settings are stored in separate std::maps. 04:25 est31 yes 04:25 est31 I know that 04:25 est31 ah 04:25 est31 thanks for reminding 04:42 est31 ShadowNinja, seen my answers on your sec-2 branch? 04:43 est31 also watch out when you merge, the " Optimize minetest.get_(all)_craft_recipe(s) " shouldn't go in there without proper approval process and so on. 04:43 est31 s/there/master 04:45 ShadowNinja est31: 1. yep, but can't fix it now. 2. will remove. 05:04 est31 ShadowNinja, PR better? 05:08 ShadowNinja est31: It would be nice if clearDefaults only cleared the defaults. ;-) 05:10 est31 ShadowNinja, yes ofc :O 05:12 est31 ShadowNinja, check again 05:16 ShadowNinja est31: Looks good, I'd just add a comment explaining why clearing are re-setting the defaults is necessary. I can't merge it ATM though (maybe tomorrow). 05:16 ShadowNinja s/are/and/ 05:16 est31 ok 06:13 VanessaE https://github.com/minetest/minetest/issues/2484 06:20 Zeno` it would appear that something is broken 06:23 VanessaE clearly :P 07:50 nrzkt well est31, other channels of distributions can be good 07:54 nrzkt wtf this night, some have drunk ? 07:55 * VanessaE looks at nrzkt 07:55 nrzkt Hi VanessaE, your crash is very... difficult :) 07:56 VanessaE yep, that's a new one on me :) 07:56 VanessaE not sure exactly what commit the server is running, just some post-0.4.12 dev build (the admin never answered my question) 07:57 nrzkt it comes at launch, right ? 07:57 VanessaE happens at connect to the server. 07:57 VanessaE connect -> get node defs, items, media -> grey window + wield + HUD -> hang 07:57 VanessaE (and sometimes it exits/crashes, perhaps with an Ubuntu crash report dialog) 07:57 nrzkt ouch 07:58 nrzkt i forget to push the isDead modification yesterday, i do it now 07:58 VanessaE doesn't seem to have a problem connecting to my servers though 07:58 VanessaE just that guy's test server. 07:58 VanessaE *shrug* 07:58 nrzkt having his last commit can be good 08:01 nrzkt ok, yesterday i thought about network and i think it may be possible to have a compat layer, but i think we only must keep it for 1 year. This layer can be used in parallel with current layer, but i need to do some tricks for brokens packets 08:01 VanessaE I can't argue with that idea. 08:02 nrzkt in devel everything is possible, but the design can be a little complicated, well the current design is complicated, then it can be possible to do something :) 08:02 hmmmm thank you. 08:02 nrzkt the code will be a little bit more complicated for the period we need to have dual stack, and tell the users to move to the new layer 08:02 hmmmm this is all i was asking for 08:03 nrzkt but the session layer must be reworked to handle the dual stack it's the touchy way. 08:03 nrzkt i don't do that for you, i do that for users. 08:03 nrzkt the more important for me is users :) 08:04 nrzkt don't forget we are working for our users. I agree loosing players is not the solution. 08:04 hmmmm the users' concerns should always come first over some big "cleanup" agenda 08:05 nrzkt Then i thought to do this on the code to have compat layer: 1. add new packet opcodes with new handlers for broken packets, to have proper packets, yes this duplicate code, but this code doesn't move. Having a duplicate is possible 08:06 nrzkt 2. add the new socket layer 08:06 hmmmm i don't see why it would need to be a layer though 08:06 nrzkt 3. integrate the socket layer in the packet queue and the session interface (i think i will rewrite all clientiface to a new session layer 08:06 hmmmm you mentioned warcraft 3 before, didn't you? so you're familiar with the warcraft 3 protocol 08:06 nrzkt non, world of warcraft 08:06 nrzkt no* 08:06 hmmmm oh 08:07 nrzkt it's a bit more complicated :p 08:08 nrzkt and the world of warcraft auth layer is very interesting too :) 08:08 hmmmm i don't know anything about WoW, but with most other blizard products, they handle backwards compatibility by just leaving support for the legacy packets 08:08 hmmmm and using new versions of the packets 08:08 nrzkt world of warcraft doesn't do this, because servers and clients upgrade together 08:09 hmmmm i don't think that's ultimately a sustainable development model when there are 0.4.10 clients still regularly being used 08:09 nrzkt the only stable API is the auth API, but their connect/auth packets are more and more proper than our :p 08:10 hmmmm so maybe instead, newer versions could use TOCLIENT_INIT_EX or something 08:10 nrzkt if the android store works properly we will have more stats. Maybe add stats on master server (i don't know if there are stats) and add a fucntion to send stats anonymously from servers can be good, no ? 08:10 hmmmm we tried that and people complained it would reduce privacy. 08:10 nrzkt i thought about TOCLIENT_INIT for new packet and rename old TOCLIENT_INIT_LEGACY 08:10 nrzkt with an option ? like debian does in its installer 08:10 est31 either way, there are already stats sent 08:11 hmmmm I don't really care, the users have to like it though 08:11 est31 its publicly known which nick plays at which server 08:11 nrzkt oh, really ? But it's not the most important, which version :) 08:12 est31 I guess we just kindly ask server owners 08:12 est31 and hope its representative 08:12 nrzkt if we add an option to server: enable_send_versionstats_to_masterserver (default: false) ? 08:13 Hunterz why is moving items between player inventory and chest so slow (laggy) in the multiplayer? 08:14 nrzkt i think it's due to the design of inventory 08:14 est31 nrzkt, also, the HTTP GET requests for the master list include an user agent 08:14 est31 so ask sfan5 for the logs :) 08:15 nrzkt oh, this can be very very good :) 08:17 nrzkt then, if everybody agree, i will add the new opcode for repaired packet to master this afternoon and close the dev-0.5 branch. 08:23 nrzkt this thing will not be used, then no problem for merge :) 08:24 est31 gtg 08:54 nrzkt this afternoon after merging the new packets (and renaming the old packets to legacy, which doesn't have any impact on handling because opcode will not be modified), i will work, if i get time, on android cmake integration 10:48 neoascetic Can this be done in foreach loop for each directory in "games/" https://github.com/minetest/minetest/blob/master/CMakeLists.txt#L148-L154 in order to make it easier to package all included games? 11:10 neoascetic Here what I am talked about: #2485 11:10 ShadowBot https://github.com/minetest/minetest/issues/2485 -- Simplify custom games packaging by neoascetic 12:08 srifqi nrzkt 12:11 srifqi neoascetic 12:11 neoascetic wut 12:19 nrzkt I'm starting the network new packets for compat layer now. I will tell you before pushing to master. This will introduce the fixed packets using new opcodes 12:28 srifqi neoascetic, have you laptop? 12:28 neoascetic Yes 12:40 nrzkt ok i tested the functions and packet rename, it's good. I push it now 13:04 fz72 Can someone please remove the rebase needed label on #2382 13:04 ShadowBot https://github.com/minetest/minetest/issues/2382 -- Remap Mouse and Key Buttons by fz72 13:06 nrzkt done 13:07 fz72 thanks 13:12 neoascetic Hooray! Night builds for OSX! :) https://github.com/neoascetic/minetest/releases 13:12 Calinou very nice 13:13 Calinou what OS X version does this require? 13:13 nrzkt it's not night here :p 13:13 Calinou you meant nightly builds 13:13 neoascetic With help of Travis-CI. I still applying #2485 and #2471 13:13 ShadowBot https://github.com/minetest/minetest/issues/2485 -- Simplify custom games packaging by neoascetic 13:13 ShadowBot https://github.com/minetest/minetest/issues/2471 -- Standalone bundle for OSX (w/ dependencies!) by neoascetic 13:13 neoascetic Not yet, first one was a day build in fact :) 13:14 neoascetic but I say my cron to build in midnight 13:14 nrzkt please squash 2471 13:15 neoascetic nrzkt done! 13:16 nrzkt i cannot test it for now i'm working on network protocol version 25 integration on master with compat :) 13:17 neoascetic okay... <_< 13:22 neoascetic Calinou it is built on Maverics so I think 10.9 and up 13:35 nrzkt okay i will push the compat layer for two client side packets now 14:04 neoascetic whom this account https://twitter.com/MinetestCommune belongs to? 14:07 sfan5 neoascetic: hoodedice 14:14 fz72 Hi Zeno` can you review #2382 again? 14:14 ShadowBot https://github.com/minetest/minetest/issues/2382 -- Remap Mouse and Key Buttons by fz72 14:19 kilbith fz72, fix the whitespaces (git rebase --whitespace=fix ) 14:24 Zeno` fz72, you've tested it? 14:27 fz72 I tested it on linux 14:32 Zeno` I think it looks ok, but I'd like to review it properly when I am less tired (12:30AM here) 14:33 Zeno` https://github.com/minetest/minetest/pull/2382/files#diff-b5b39c3a29359759ac6f0293dc37a453R305 14:33 Zeno` ^--- I just hope that isn't a performance killer 14:35 Zeno` perhaps getKeySetting("keymap_dig") should be cached in the class somehow (making sure it gets updated if the user changes key bindings of course) 14:35 Zeno` std::map lookups are not exactly fast and this happens every loop 14:36 Zeno` grr "low priority" label 14:37 Zeno` everything is low priority heh 14:37 Zeno` most useless label ever created :D 14:37 fz72 kilbith: done Zeno`: Random InputHandler can maybe done in a other way. 14:39 kilbith hmm, the whitespaces are still here 14:39 fz72 The lookups in std::map are done for all buttons the same way. 14:40 fz72 kilbith: Should I do the rebase in the same branch or in a new one? 14:41 kilbith same branch 14:43 Zeno` fz72, overall I think it looks fine. But I'll review more properly tomorrow. I.e. at a glance it looks mostly good (a few minor issues -- or potential ones -- but nothing major) 14:44 Zeno` fz72, I don't think the keys are looked up using the getKeySetting() function (I recall removing that) 14:44 Zeno` s/removing/replacing 14:45 fz72 You mean in RandomInputHandler? 14:46 fz72 In RealInputHandler the keys are cached 14:47 Zeno` game.cpp 14:47 Zeno` https://github.com/minetest/minetest/pull/2382/files#diff-18513665750ef5adf42b5ec29e14162eR3009 14:47 Zeno` is effectively using the std::map rather than the key cache 14:48 Zeno` oh wait 14:49 Zeno` my apologies 14:50 fz72 no problem ;) 14:51 Zeno` ok, well it looks good 14:51 Zeno` https://github.com/minetest/minetest/pull/2382/files#diff-b5b39c3a29359759ac6f0293dc37a453R117 14:51 Zeno` else if should be on the same line as the closing brace of the previous 14:51 Zeno` } else if { 14:52 Zeno` minor style issue, otherwise I'm ok with it 14:52 Zeno` nice ;) 14:53 Zeno` same with else... 14:53 Zeno` } else { 14:53 Zeno` ... 14:53 fz72 there are much else and else if... 14:54 Zeno` only bother with those in inputhandler.h and I think it can be merged 14:55 Zeno` (yes, this is the pedantic part of it, :( 14:56 Zeno` I'll +1 it now though 14:57 fz72 I'll fix the else and else if later 14:57 Zeno` ok, I've give my approval already so you just need another core dev to approve as well 14:58 Zeno` if it's not merged soon I can chase it up 14:59 Zeno` I'll remove the low priority label also 15:09 fz72 Zeno`: I fixed else and else if. 15:10 fz72 kilbith: I did the whitespace=fix as you said, but it seems that it doesn't change anything 15:10 kilbith odd 16:41 neoascetic Hey, could someone put a link on https://github.com/neoascetic/minetest/releases as a possible way to install minetest on OSX on minetest.net? 16:48 nrzkt how this is generated ? 16:49 neoascetic cron on my server + travis-ci-based deploy 16:49 neoascetic latest each 30 minutes, night - every night 17:46 nrzkt neoascetic you build with which OS ? 17:49 neoascetic nrzkt OSX Macericks on travis-ci (yep, it supports osx) 17:51 Calinou what OS X version does this require? 10.8? 17:52 neoascetic I don't know exactly, it works on my 10.9.5 :) 17:53 Calinou the current latest is 10.10 17:53 neoascetic yes, I know 17:53 neoascetic it works on 10.10 17:53 neoascetic I've checked 17:56 neoascetic what you all think about #2485? 17:56 ShadowBot https://github.com/minetest/minetest/issues/2485 -- Simplify custom games packaging by neoascetic 18:09 nrzkt we support android 2.3... i don't think android 2.3 phones can run MT properly xD 18:33 nrzkt ok guys, i'm now able to test MT on my phone :D 18:34 nrzkt wtf MT wants to access to my contacts 18:34 rubenwardy Is that a real build? 18:35 rubenwardy Or is it from play? 18:35 nrzkt build on my computer with my android sdk and ndk :p 18:35 nrzkt and if i works i could prepare a release engineering for MT on play store 18:38 nrzkt and i could work on the app easily :) 18:38 Calinou I wonder if we could port minetestmapper 18:38 Calinou using a file manager, you could map worlds from your device 18:38 Calinou mobile phone is good for exploring maps 18:41 nrzkt the MT controls on a phone (and i have a Huawei Ascend Mate, a phablet), are... ugly :s 18:42 nrzkt having a button on the first finger to interact is better than interact with the screen itself :s 18:43 nrzkt the build process is not very clear... MT added my installed mods from my machine to the apk 18:45 nrzkt i will prepare an MT release for 0.4.12 from stable-0.4 branch this evening and prepare the play store 18:45 nrzkt i think we must add a special directory to use for google play store description translations, no ? 18:45 rubenwardy I have a Huawei Y300, it's sh_t 18:47 rubenwardy I think it is a good idea to have an official presence on the play store. Some members have ethical issues with the £15 required to start an account, but meh. It means you get more testers as it's easier to get and you don't risk spyware. 18:48 nrzkt right, and we can get real stats of MT versions :) 18:48 nrzkt and i already have an account for my own devels, out of MT, then i can use it for MT, not a problem :) 19:03 celeron55 i do have an account too 19:03 celeron55 but as i try to focus on other things i don't have a preference on who's account is used 19:04 nrzkt then who release ? i can do it as i follow the commits and i can provide a hotfix to play store faster than you, no ? :) 19:22 nrzkt ok, i finished the new unused handlers for protocol v25 init/auth/password, which permit compat on serverside for v24 and v25 protocols. I will push it in some minutes. This protocol couldn't be used with this commit because the auth mechanism will not be implemented yet, only the legacy auth is used. the next step will be a cleanup on packet sending, backported from dev-0.5 19:37 est31 nrzkt, do you know that srp needs more than 2 messages? 19:41 nrzkt this is a first design, and i haven't implemented SRP, this code is for compat 19:41 nrzkt maybe a defintive 0.4.13 19:41 nrzkt with the two network layers but not the break 19:49 nrzkt this design split the legacy _Init packet into _Init and _Auth, this permit to have a stable _Init packet for the future :) 19:50 est31 ah 19:50 est31 init packet could contain the list of supported auth mechanisms 19:50 est31 like ed25519 or srp 19:51 est31 or no bad idea 19:51 nrzkt hmmm this isn't bad but difficult to handle 19:56 nrzkt i think we can add a field now for a future use, that could be useful, but we mustn't misuse it :( 20:15 est31 No bad idea, init packets should be as small as possible 20:16 nrzkt right 20:59 hmmmm celeron55: remember how we were talking about pseudorandom generators a while ago? between that time and now I found this PCG which looks REALLY good 20:59 hmmmm http://www.pcg-random.org/download.html 21:00 hmmmm convince me why I shouldn't copy paste that code in the first paragraph of the website into PseudoRandom right now. 21:00 nrzkt i will push another backport from dev-0.5 branch. Remove the use of pointers for NetworkPackets and use references, as pointed by ShadowNinja 21:01 hmmmm (well, aside from the breakage in pseudorandom structures :P) 21:02 neoascetic I hope #2471 will be merged before 0.5 release 21:02 ShadowBot https://github.com/minetest/minetest/issues/2471 -- Standalone bundle for OSX (w/ dependencies!) by neoascetic 21:02 hmmmm in re references: I think we should make it a code style rule to forbid references in parameters unless they're const 21:02 hmmmm that would eliminate the reason why I'm not a fan 21:03 nrzkt i'm not fan too, by it's generally the good approach for this. And i haven't made them const. You are right, they should be because packets mustn't be modified outside of the generators, i modify the Send() functions in consequence 21:04 hmmmm nrzkt: looking at your actual commit, this doesn't seem to be what I'm talking about 21:04 nrzkt oh :) okay 21:05 hmmmm should Client::Send(NetworkPacket *pkt) be Client::Send(const NetworkPacket *pkt) ? 21:05 nrzkt i think it could be good but the compiler doesn't like it :) 21:05 hmmmm well 21:06 nrzkt and it's a pointer, then it's not applicable, i'm right, i need to have Send(const NetworkPacket pkt) instead, right ? 21:06 nrzkt s/i'm right/if i'm right/ 21:09 nrzkt i will also push now two typo fixes 21:09 hmmmm huh? no, but you'd need to make Connection::Send() be const-correct too 21:09 nrzkt a comment and a break after assert which is missing and some compilers doesn't tolerate it and warn 21:09 hmmmm const is viral in that sense 21:10 nrzkt what would you mean ? 21:13 nrzkt hmm okay i understand what you mean. I'll look at this case later 21:13 hmmmm in any case, I keep seeing NetworkPacket(THE_PACKET_ID, 0, peer_id) ... the 0 seems kind of redundant 21:13 hmmmm why not have the reserved data size come after the peer id 21:13 hmmmm so that way you can make the parameter optional 21:14 hmmmm NetworkPacket(u16 command, u16 peer_id, u32 datasize=0) 21:14 hmmmm if (datasize) m_data.resize(datasize); 21:14 nrzkt a legacy choice in fact, nobody comment it. And i already have a NetworkPacket(u16 command, u32 datasize); 21:15 nrzkt then this will conflict :( 21:16 hmmmm oh. yeah I see it 21:16 hmmmm is m_data.clear() necessary in the dtor? 21:17 nrzkt i do it generally when i do c++, to be sure. maybe a more specialized official documentation can prevent this call 21:19 nrzkt right for the m_datasize = 0 and resize on the constructor, i will add the test 21:24 hmmmm well i don't think it makes a difference aside from setting an internal variable... 21:25 nrzkt no, i choose only to use same variable because we read it just before, to get the same way. The test was here, it's the more important 21:27 nrzkt most* 21:33 nrzkt i must go, good night all :) i will continue the backport this week-end