Time Nick Message 06:42 celeron55 proller: no no no 06:42 celeron55 no 06:42 celeron55 and no 06:43 celeron55 half of that doesn't make any sense 06:46 celeron55 or not half but... some 06:46 celeron55 the addition of ", used for client-server transfer" to serialization.h is, like, poisonous 06:47 celeron55 it's wrong documentation; version 26 is a HACK you're doing, not something that will be endorsed in the long term 06:49 celeron55 then, why do you do "int version = protocol_version >= 21 ? 2 : 1;" in NodeBox::serialize()? 06:49 celeron55 oh, actually that's right 06:49 * celeron55 is waking up 06:51 celeron55 ugh 06:54 celeron55 i really want serialize.h to read "26: Never written; read the same as 25", and have SER_FMT_VER_HIGHEST to be changed to SER_FMT_VER_HIGHEST_READ and SER_FMT_VER_DISK to be changed to SER_FMT_VER_HIGHEST_WRITE... and then add MapBlock::serializeNetworkSpecific() which will be added separately to the TOCLIENT_BLOCKDATA packet 06:54 celeron55 it's not clean otherwise 06:55 celeron55 and remove everything you have ever added to MapBlock::serialize/deSerialize 07:08 hmmmmm why did this mess happen in the first place 07:09 hmmmmm am i mistaken, or is what this is accomplishing completely trivial 07:09 hmmmmm i have no clue why you can't just have one mapblock serialization version for both disk and network 07:10 hmmmmm er, rather, why this isn't handled in the same manner as things that are already not transferred over the network 07:23 celeron55 hmmmmm: you're not being too clear about how you think it should be made 07:23 celeron55 what are those "things that are already not transferred over the network"? 07:24 celeron55 this is a thing that *is* transferred over the network, but *not* saved on disk; which didn't yet exist and because of that proller did it terribly and nobody didn't notice why it was wrong 07:25 hmmmmm maybe i should read up on some past conversations in case i missed something 07:25 celeron55 (also that patchset combines fixes for two completely separate things) 07:25 celeron55 hmmmmm: i want you to do a thing ASAP, by the way 07:25 celeron55 hmmmmm: you need to rename your fork of jthread NOW 07:26 hmmmmm but this is beside the point, why isn't the entire mapblock version at 26 07:26 celeron55 it's causing real packaging problems 07:26 hmmmmm on some odd linux distros... it should be enough to just disable searching for the system copy 07:26 hmmmmm do you actually want me to *rename* it? how do i even go about that 07:27 celeron55 odd linux distros? hell, all linux distros provide their own version of jthread 07:27 celeron55 it's not odd in any way 07:27 hmmmmm i don't recall mainstream distros having problems with this 07:27 celeron55 you have to rename it because libraries are searched by name 07:27 hmmmmm just this one that nobody's ever heard of 07:27 hmmmmm and that was a long time ago but nobody seemed concerned about it for quite a while 07:28 hmmmmm so tell me exactly what i'd need to do here (i'm not too knowledgable with cmake) 07:28 celeron55 isn't it enough for you that having a different library with the same name is just totally wrong? 07:29 celeron55 well me neither 07:29 hmmmmm i'm guessing here but, first i'd need to remove cmake/Modules/FindJthread.cmake 07:29 hmmmmm who is the one who originally did the cmake build system anyway? 07:30 celeron55 me, altough it contains more lines by others than me 07:30 hmmmmm alright hold on a minute, i don't see what's wrong with just having it always use the project version 07:30 hmmmmm i'd omit the majority of FindJthread.cmake 07:31 hmmmmm just leave those three statements after ELSE(JTHREAD_FOUND) 07:31 hmmmmm that would fix it, sure, but it's not really a complete solution 07:31 celeron55 umm 07:31 hmmmmm renaming stuff is a lot of work you know 07:31 celeron55 so does that fix the linker getting the global jthread via -ljthread? 07:32 celeron55 also, does that make cmake not use a global FindJthread if such exist? 07:32 hmmmmm it does?? well shit 07:32 celeron55 i don't know 07:32 hmmmmm i'd rather look it up and be sure though 07:33 hmmmmm but nevermind this 07:33 hmmmmm i wouldn't do something as hacky as that, renaming it would be the proper thing 07:34 celeron55 i'd like to have Akien now here trying stuff out 07:34 celeron55 maybe he comes if we wait a bit 07:34 hmmmmm akien?? 07:34 hmmmmm oh, i guess he was the most recent person to complain about this 07:34 celeron55 that's who complained about it last evening 07:34 hmmmmm sorry, haven't been keeping up with things 07:35 hmmmmm you realize i won't be able to actually test this myself, right? 07:35 celeron55 yes, and me neither 07:36 celeron55 let's wait for Akien, i don't like and don't expect anyone else to like doing unnecessary work... 07:37 celeron55 it could be that this can be solved in some very simple way but we can't know unless it can be tested 07:37 hmmmmm i guess this is just three cases actually: me, whose os doesn't have a jthread package in the first place, someone on linux who has it available but uses the project version instead, and then someone who has the actual jthread globally installed 07:37 hmmmmm well like i said 07:37 hmmmmm the very simple way would be to futz with the FindJthread.cmake file 07:37 kahrl if we are going to spend so much time on this issue, why don't we get rid of all the modification and make it possible to use the system jthread again? 07:37 hmmmmm because jthread is just shit 07:37 celeron55 because the system jthread is so featureless 07:38 hmmmmm everything i added to jthread is completely necessary 07:38 kahrl the modifications can be moved to porting.h 07:38 hmmmmm the IsSameThread? 07:38 celeron55 no they can't because they are private stuff in the classes 07:38 hmmmmm the semaphore stuff? 07:38 kahrl oh, IsSameThread isn't even needed, see threadid_t in threads.h 07:39 hmmmmm erm 07:39 hmmmmm i don't recall all the exact details about this, but i did it this way for a specific reason 07:39 hmmmmm it's almost 4 am so i'd rather not get into it right now 07:48 Akien celeron55, hmmmmm: I'm there now :) 07:48 Akien Just read the backlog 07:49 celeron55 do you have the non-building setup at hand now? 07:50 Akien Yep 07:50 celeron55 so, i want to know a few things before continuing 07:50 celeron55 do you have a global jthread installed? 07:51 Akien I do, 07:51 Akien And it's dev headers 07:51 celeron55 have you modified minetest's cmake/Modules/FindJthread.cmake? 07:52 celeron55 or any of the other build scripts 07:52 Akien Currently I'm trying to build a clean release of minetest 0.4.7 and minetest-game 0.4.7 07:53 Akien I removed our patches for 0.4.3 to have it use the system jthread 07:53 Akien If I try to build simply with cmake && make -j2, it works ok. 07:53 Akien My problem happens when using the shared library instead of the static one. 07:53 celeron55 and the errors you pasted yesterday are what you get now? 07:54 celeron55 oh, what works? 07:54 Akien I mean, if I build a local copy for me with a simple cmake it works. But to package for Mageia, I have to follow some specific policies, such as IINM using shared libraries instead of static ones. 07:55 Akien So I try to configure with "-DBUILD_SHARED_LIBS:BOOL=ON" in the cmake instruction (among others, flags and so on) 07:56 Akien Then the errors occur 07:56 celeron55 okay, and that will attempt to build our fork of jthread as a shared library; the problem you ran before you left yesterday was the fact that our jthread's build system is broken in the first place for a shared build 07:56 celeron55 if those errors are fixed, you are going to stumble upon the problem that it's a different library with the same name as what you already have in the system 07:57 Akien Well I guess if it's bundled in the minetest data directory, it should not interfere with the system jthread, would it? 07:57 celeron55 our request for everyone has been that simply don't try to build it as a shared library, it's not useful as such 07:58 celeron55 if you build it as a shared library, of course it will (cmake will install it in a global location) 07:59 celeron55 so yeah, kahrl: what do you think? 07:59 kahrl uhh, I'm not even sure anymore 07:59 kahrl I don't see the private things you were talking about 08:00 Akien I'll check the policy with some advanced Mageia packagers, it might be that including it as a static library is the way to do it. 08:01 celeron55 Akien: well, we now know the exact issue and we'll try to get it addressed sometime; as for now, you know what to do 08:02 celeron55 kahrl: modifying jthread was a bad idea right from the start... but i'm not going to start reworking it 08:03 celeron55 if you're right about there being nothing to stop it, it's probably quite a little work 08:03 kahrl I'm doing it right now 08:03 kahrl it's not that bad actually 08:07 celeron55 i wonder why there isn't an issue with lua... oh heh, we have conveniently forced a static build in it's cmake configuration 08:08 celeron55 distro zealots can't cause it to break with their silly flags 08:08 celeron55 :-D 08:09 celeron55 we could add a single keyword in jthread's scripts and nobody would ever complain again 08:11 celeron55 all this jthread stuff could really just be considered to be a bug in minetest's build system 09:13 sfan5 Can someone verify that this builds correctly? https://github.com/sfan5/minetest/tree/security 09:22 Akien I patched (locally) minetest-0.4.7 as follows to fix the locale path on UNIX: 09:22 Akien - set(LOCALEDIR "${CMAKE_INSTALL_PREFIX}/share/${PROJECT_NAME}/locale") 09:22 Akien + set(LOCALEDIR "${CMAKE_INSTALL_PREFIX}/share/locale") 09:23 Akien I don't know if it's Mageia specific or if most modern distros use the same policy, but all locales go in /usr/share/locale//.mo 09:24 Akien (I forgot to say it was in minetest-0.4.7/CMakeLists.txt 09:52 Akien Building with static libraries, minetest works great. 09:53 Akien You all did awesome work, I'm impressed :) 09:53 Akien With all visual effects disabled (my laptop is reaaaaally old), minetest runs really smoothly compared to minecraft. 10:04 kahrl so I reverted all custom changes to jthread 10:04 kahrl https://github.com/kahrl/minetest/commit/681d6df31b672a92cd3e1618668279ad93a6cddc 10:05 kahrl replaced the IsSameThread based implementation of getCurrentMapgen with a TLS based solution 10:05 celeron55 you should leave in "// CriticalSection is way faster than the alternative" 10:05 celeron55 otherwise nobody understands why the #define is there (it's not in the jthread distribution either) 10:05 kahrl I'll do that 10:06 kahrl should I update jthread to 1.3.1? 10:06 celeron55 or, actually 10:06 celeron55 a better alternative would be to define JMUTEX_CRITICALSECTION in cmake 10:06 kahrl I would put that comment in jthreadconfig.h then 10:07 celeron55 umm 10:07 kahrl another question is whether the revert of the _WIN32_WINNT change breaks the win32 build 10:07 celeron55 can you make sure that both versions of jthread are supported? 10:07 celeron55 because, like, we're likely to see both 10:07 kahrl let me see 10:08 kahrl the biggest change in 1.3.1 is that is uses a namespace jthread 10:08 celeron55 yes; i haven't looked more than that 10:09 kahrl "using namespace jthread;" should work with both, right? 10:09 celeron55 i'd guess so 10:09 celeron55 Akien: what is the patch that you have used that makes minetest use system jthread? 10:11 celeron55 because 0.4.3 should already do that by itself; you having a patch means that it's broken 10:13 Akien celeron55: http://svnweb.mageia.org/packages/cauldron/minetest/current/SOURCES/minetest-0.4.3-jthread.patch?view=markup 10:14 celeron55 oh, so it's just about supporting jthread 1.3 10:14 Akien I did not create it though, yesterday was the first time I looked at minetest's source 10:15 Akien Yes we use jthread 1.3.1 10:39 PilzAdam kahrl_, your jthread patch does not build with mingw: /home/adam/Minetest/win32/minetest/src/porting.cpp:174: error: ‘OpenThread’ was not declared in this scope 10:43 kahrl_ ah, glad you caught that 10:43 kahrl_ must be because of the _WIN32_WINNT thing? 10:43 PilzAdam I think so 10:43 celeron55 just throw that in porting.h i think 10:44 celeron55 ...or maybe threads.h? 8) 10:44 kahrl_ I've already changed it so the only jthread includes are in threads.h and util/thread.h 10:44 kahrl_ so yeah I'll do it in threads.h 10:45 kahrl_ pushed https://github.com/kahrl/minetest/commit/779272c6f4d2db21d7e25aea16352fb701c810b4 10:47 kahrl_ bbiab 10:51 PilzAdam that builds correctly 12:00 kahrl_ other than the namespace change I don't see anything in jthread 1.3.1 that we would have to do something about 12:01 kahrl_ (except that JMUTEX_CRITICALSECTION is now called JTHREAD_CONFIG_JMUTEXCRITICALSECTION) 12:01 celeron55 i still wonder why jthread doesn't default to using that 12:02 celeron55 oh now i remember 12:02 celeron55 or, well... the functional difference in addition to some process/thread stuff (which doesn't generally matter) is that JMUTEX_CRITICALSECTION is a recursive mutex 12:03 celeron55 compared to the alternative, it's faster by an order of magnitude though 12:03 celeron55 (while still slower than pthread) 12:24 sfan5 celeron55: any comments? http://pastie.org/pastes/8196563/text 12:25 sfan5 wait.. 12:25 sfan5 forgot the "[Man-in-the-Middle is also no possible]" thing 12:25 sfan5 *forget 12:26 PilzAdam how is man-in-the-middle not possible here? 12:27 sfan5 ..I already revoked my statement of it not being possible 12:36 proller celeron55, like this ? https://github.com/proller/minetest/compare/ver 12:36 * sfan5 thinks celer*n55 is away 12:47 kahrl_ updated jthread to 1.3.1: https://github.com/kahrl/minetest/commit/c1e64cf3f91c5e226eca8c1b82771ad0778a2a30 12:47 kahrl_ all the headers and sources are straight from jthread-1.3.1.tar.gz 12:47 kahrl_ CMakeLists.txt is modified a little 12:55 PilzAdam kahrl_, tested with gcc and mingw 12:58 kahrl_ I tried it with jthread 1.3.1 installed by gentoo's portage, compiles and works 12:58 kahrl_ (once I pass -D JTHREAD_INCLUDE_DIR=/usr/include/jthread to cmake) 13:00 kahrl_ PilzAdam: does it set JTHREAD_WIN32_CRITICALSECTION=ON in jthreadconfig.h? 13:02 kahrl_ err, I mean "#define JTHREAD_CONFIG_JMUTEXCRITICALSECTION" 13:02 PilzAdam JTHREAD_WIN32_CRITICALSECTION:BOOL=ON 13:03 PilzAdam yea, it sets it 13:03 kahrl_ ok good 13:37 kahrl_ should I add search paths to FindJthread.cmake, so that it automatically finds the system jthread on gentoo (and other distros that don't put jthread.h into /usr/include)? 13:37 kahrl_ like https://gist.github.com/kahrl/6131384 13:38 Akien kahrl_: /usr/include/jthread/ should work on Mageia (and at least on Fedora I think) 13:39 Akien In most modern distros actually 13:39 Akien s/In/On/ 13:40 kahrl_ well, the build system of jthread puts it into /include/jthread by default 14:37 proller melting-freezing cause new ice have lower light than before 14:48 BlockMen warum ist in nodedef.cpp#L218 drowning = true ? 14:48 BlockMen ups..wrong window^^ 14:53 sfan5 BlockMen: damit man ohne noclip in blöcken "ertrinkt" (soweit ich weiß) 14:55 RealBadAngel about drowning, we shall allow this attribute for airlike nodes too 14:56 RealBadAngel not only liquids 14:56 RealBadAngel this will allow to create outer space 14:57 sfan5 you mean outer 0x20 14:58 RealBadAngel i mean worlds without air 14:58 RealBadAngel thus no breathable atmosphere 14:59 RealBadAngel look at paramat's worlds 15:01 RealBadAngel this with my latest patch could allow space suits 15:08 BlockMen RealBadAngel, working on that (too) :P 15:10 proller RealBadAngel, and underwater suits ? 15:11 proller and now possible to make cold-hot damage and wear ? 15:13 RealBadAngel underwater are already possible 15:13 RealBadAngel i will add one to technic 15:14 RealBadAngel i made one for testing the patch out of flashlight code 15:15 RealBadAngel i just put there check if breath is less than 5 and set it to 10 again 15:15 RealBadAngel so, piece of cake 15:16 RealBadAngel one thing that should be changed is just if check. it looks now for liquid and drowning attribute 15:17 RealBadAngel it shall be changed to check only for drowning 15:17 RealBadAngel and then let other than liquids nodes carry that attribute 15:18 RealBadAngel also at the very same place suffocation shall be added 15:19 RealBadAngel we do already check node 15:20 RealBadAngel so if solid, suffocate, if liquid or airlike without breathable attribute -> drowning 15:26 BlockMen RealBadAngel, plz wait before patching, im changing drowning atm (that you can modify damage of drowning) 15:27 RealBadAngel that was just the ideas 15:27 RealBadAngel im too busy with shaders atm to work on anything else 15:27 RealBadAngel i plan to finish the current stage of shaders changes this weekend 15:28 RealBadAngel then i will go for technic coding for a week 15:28 BlockMen ok. im nearly done with that, is it ok when i do that with the supported node types too then? 15:29 RealBadAngel imho you shall drop function in_liquid or something like this 15:30 RealBadAngel and just check for drowning attribute 15:30 RealBadAngel this alone will allow to create void or atmospheres 15:30 RealBadAngel or even poisonous gases on swamps 15:31 RealBadAngel for example 15:32 jin_xi miasma 15:33 RealBadAngel wut? 15:34 BlockMen RBA, atm all nodes have drowning attribute, so that needs to be changed then too 15:37 RealBadAngel its good then 15:37 RealBadAngel just the check has to be changed 15:37 RealBadAngel i mean check if it is a luquid 15:37 RealBadAngel *liquid 15:40 RealBadAngel https://github.com/minetest/minetest/blob/master/src/environment.cpp#L2245 16:16 RealBadAngel ok, folks i think i can spread the news by now 16:17 RealBadAngel i just got a confirmation 16:17 RealBadAngel next month on biggest polish PC gaming magazine CD Action, there will be review of our game 16:18 RealBadAngel we need to prepare stable releases for cover CD for the magazine 16:18 Jordach we need 0.4.8 NOW. 16:18 RealBadAngel because game is gonna be shipped with the magazine 16:19 RealBadAngel we need all the possible and rock solid builds for this 16:20 RealBadAngel that means 30k of copies and possible new users 16:20 proller ups 16:21 proller push to topic 16:21 proller with dates 16:22 Jordach RealBadAngel, we will need a windows installer 16:22 sfan5 RealBadAngel: we have a polish translation, right? 16:22 sfan5 Jordach: we have one 16:22 RealBadAngel yup i made one 16:23 RealBadAngel but the installer would be nice 16:23 RealBadAngel lets make it kiss as much as possible 16:23 sfan5 SFX? 16:24 RealBadAngel dunno, im not familiar with installers 16:24 * sfan5 can make SFX packages 16:24 RealBadAngel so you have been chosen as a volunteer ;) 16:25 sfan5 but I'm not the guy who's familiar with GUI programming on Windows 16:25 sfan5 bah, I need a static zlib for windows 16:25 RealBadAngel we do have circa 2 weeks to the deadline 16:25 Jordach sfan5, NULLSOFT installer 16:26 sfan5 Jordach: not simple enough 16:26 Jordach or use ruben's script 16:26 RealBadAngel we have to send proper and checked code before 14th this month 16:26 Jordach we have 13 days 16:26 RealBadAngel then all the content goes to make cover dvd's 16:27 RealBadAngel we cannot cross this line 16:27 Jordach this looks like a good time to merge Sam II 16:27 RealBadAngel if we do, we will be delayed by month 16:27 RealBadAngel Jordach, technic_game waits for it 16:28 Jordach starting git 16:28 proller and two pulls for mobs 16:28 proller AND MY VERSION COMPAT BUGFIX 16:28 proller after celeron55 agreed 16:29 RealBadAngel hmmmm, have you read the news? 16:30 Jordach RealBadAngel, i dont have push to technic_game 16:30 RealBadAngel ooops, lemme fix that :) 16:30 Jordach wait 16:30 Jordach i do 16:33 celeron55 it's a semi feature-freeze then 16:35 hmmmm oh no oh no 16:35 hmmmm karhl don't use TLS! 16:35 celeron55 RealBadAngel: what kind of stuff have you even told to them? 16:35 celeron55 i hope it's more about community and open source than shaders and bumpmapping 16:35 hmmmm karhl, that's C++11 16:36 Jordach celeron55, i have a texture fix waiting for merge (minetest_game #184) 16:36 RealBadAngel i told them about free game 16:36 celeron55 Jordach: find someone else to deal with that 16:36 RealBadAngel mc like 16:37 Jordach celeron55, like PilzAdam will ever do that 16:37 RealBadAngel guy have tested it on his own and liked the project 16:37 celeron55 Jordach: i have more important things on my list now 16:37 celeron55 like unbreaking proller's code 16:37 RealBadAngel so he heared my voice and agreed to make a rewiev 16:39 celeron55 proller: it's still wrong 16:39 celeron55 proller: MapBlock::deSerializeNetworkSpecific should work according to the protocol version, not map version 16:39 RealBadAngel celeron55, shaders etc are just my playground, same as v7 for hmm or game for PA 16:39 RealBadAngel i dont differ the whole project by our playgrounds at all 16:40 RealBadAngel it is our hobby and we all do our best for it 16:40 RealBadAngel thats it 16:42 RealBadAngel my gf sittin next to me is playin mt at the moment 16:42 RealBadAngel and shes screamin when i get the shaders patch done 16:43 RealBadAngel my friends in Berlin also demands updated game 16:44 RealBadAngel i think im doing good for the community 16:48 sfan5 celeron55: you have a comment about my kinda proposal for protocol encryption? 16:50 celeron55 sfan5: what's the point of it? 16:50 sfan5 the point of encryption? 16:51 hmmmm nevermind about that last concern of mine, i didn't read the patch completely 16:51 celeron55 the actual question still is more like "whether to do it or not" than "how to do it" 16:51 celeron55 i don't care about implementation until i know what it's for and who needs it 16:52 RealBadAngel celeron55, i shall read your comments as cons to the idea of magazine rewievin our project? 16:52 hmmmm from what i understand, sfan5 is afraid of somebody MITMing his traffic and converting messages to saying "heil hitler", iirc it's highly illegal to praise anything related to the nazi party in germany 16:52 celeron55 RealBadAngel: i have said no cons or pros about it 16:52 RealBadAngel so decide please 16:53 hmmmm although it's totally possible somebody else could do that with HTTP or something 16:53 sfan5 hmmmm: thats not my personal concern but could be other peoples concern 16:53 celeron55 RealBadAngel: well, i already said the main concern i had 16:53 RealBadAngel your name is going to be printed in 30k copies 16:53 sfan5 anyway, I was more asking whether anyone sees any big mistake in that proposal 16:53 hmmmm that shouldn't be a concern at all, first off the internet is fairly anonymous and your privacy's not going to get F'd over something as trivial as that 16:54 hmmmm second, if that were actually to get you in trouble and go to a trial they'd merely need to check some hops closer to your own and see you're innocent 16:54 RealBadAngel celeron55, u have said if i will be making a review 16:54 RealBadAngel thats wrong 16:54 hmmmm third, this is a ridiculous, contrieved scenario which will never and has never happened in real life 16:55 RealBadAngel guy from the magazine will judge us 16:55 hmmmm I vote against encryption 16:55 hmmmm it's just.... really not necessary 16:55 RealBadAngel so pointing shaders stuff was meaningless 16:56 sfan5 hmmmm: it could be optional 16:56 celeron55 RealBadAngel: well, articles about stuff generally have some "point" in them; i just wonder what that point will be about minetest 16:56 hmmmm yes, but nevermind the fact you're adding all this extra stuff in that nobody's using but yourself 16:56 RealBadAngel he will judge You, me, hmm and everybody else for the job done 16:56 Calinou there is encryption in xonotic: noone uses it 16:56 celeron55 i vote for encryption to be considered when the protocol is redesigned more thoroughly 16:57 celeron55 which will happen... ehm... well, nobody knows, but probably some year 16:57 hmmmm i bet xontonic also uses SRP for logon because of some paranoid security fear 16:57 RealBadAngel celeron55, i have no idea what will be the point, but one thing im sure it will be good for us 16:57 hmmmm as for the magazine thing, i personally don't like publicity so i'd rather stay out of it 16:58 Calinou publicity is always a good thing 16:58 RealBadAngel :) 16:58 hmmmm i really wonder if minetest needs more contributors or not too, i mean too many cooks :/ 16:58 Calinou we don't have enough already 16:58 BlockMen celeron55, what you think about my pull on template? 16:58 RealBadAngel it means bigger community 16:58 RealBadAngel and thats good 17:00 hmmmm as for the threading thing, i'm not sure why karhl chose that route; jthread never did anything, so forking it off imo would be the smarter thing to do 17:00 hmmmm merely for organizational purposes 17:01 hmmmm i personally can't see any benefit at all to using a system jthread library 17:01 hmmmm ever, in any case 17:02 celeron55 well, none of this matters other than the fact that after what kahrl did, it works on everything directly again and it makes people happy 17:02 hmmmm when i first modified it, it was my intent to force it to use the project version always and/or rename it, but I didn't want to screw around with the build system at that time and nobody seemed to have a problem 17:03 PilzAdam so, we have a feature freeze now because of a random magazine bringing an article about Minetest? 17:03 sfan5 kind of 17:03 hmmmm i'd say we have a feature freeze because this nonsense has gone on long enough 17:03 celeron55 PilzAdam: that needs to be considered now, i think 17:03 hmmmm 0.4.8 should've been out two months ago 17:03 celeron55 do we want to put out a version? 17:03 hmmmm but 17:04 hmmmm we're not releasing anything until the weather bullshit gets fixed 100% 17:04 PilzAdam I think weather should be reverted for 0.4.8 17:04 hmmmm yeah, probably. 17:04 celeron55 the compatibility issue is going to get fixed tomorrow as proller comes along again fixes the last thing that i noticed 17:04 hmmmm if you want 0.4.8 out quick 17:04 celeron55 +and 17:04 hmmmm well tomorrow is good 17:05 hmmmm having a different version for the network is nonsensical though, i hope by fix you mean move everything to 26 17:05 PilzAdam a feature freeze now would mean that RealBadAngel's shader code has to wait 17:05 hmmmm grrrr 17:06 celeron55 hmmmm: i can assure you the fix is sane in all imaginable ways; don't worry about it 17:06 RealBadAngel i wanted to push it this weekend 17:06 RealBadAngel but it can wait 17:06 RealBadAngel no problemo 17:07 celeron55 hmmmm: i could explain it completely but i'll rather not, the logic behind versioning is kind of hard and long to explain in a thorough way 17:08 hmmmm to the people who complain about long waits between versions: I don't believe our release cycle is really that bad, considering the amount of stuff that was changed or added is really a spit in the bucket and nothing in comparison to some other FOSS projects 17:09 celeron55 one thing i've been left wondering now is the question: is game design by committee better than game design by random? 17:10 celeron55 we kind of have those options 8) 17:10 hmmmm you say "by committee" as if this is what's happening 17:11 celeron55 what's happening is somewhere in between of those 17:11 PilzAdam currently nothing is happening 17:11 hmmmm "design by committee" is a lot worse than this. like ATM packet size 17:11 PilzAdam we dont even have a game designer 17:11 RealBadAngel in your opinion PA 17:11 hmmmm the europeans wanted 32 bit packets for less latency because they wanted to use it for telephone calls 17:11 hmmmm the 'murricans wanted 64 bit packets for higher throughput for data applications 17:11 celeron55 PilzAdam: well, now i mean everything, including eg. jordach's new model 17:12 celeron55 we need a way to choose those kinds of things 17:12 hmmmm so the stupid committee decided "we should please both parties and make it 48 bytes" 17:12 hmmmm err, i meant byte, not bit before 17:12 RealBadAngel hehe 17:12 sapier1 sfan5 is there any reason for you to put work in a outdated unsafe crypto library instead of finding a prooven maintained one first? 17:12 kahrl_ hmmmm: I personally don't mind always using a modified in-project jthread (since it's so small), but the topic keeps coming up here repeatedly and the time talking about it could be used more productively 17:12 kahrl_ so I fixed it once and for all 17:12 celeron55 we have way more headroom in how much work could happen on the game content side; on the engine side we're kind of saturated 17:13 hmmmm karhl, really to 'fix' it all you'd need to do is remove some lines from cmake/Modules/FindJthread.cmake 17:13 hmmmm but that's hacky 17:13 RealBadAngel content side shall be left for game creators 17:13 kahrl_ well, distributors will keep asking how to "unbundle" it 17:13 kahrl_ so it's not a full solution 17:13 PilzAdam celeron55, I kinda stopped development of minetest_game because too many people complaining about it 17:14 celeron55 kahrl_: what do you think of adding STATIC to jthread/CMakeLists.txt:16 like there is on lua/src/CMakeLists.txt:63? 17:14 celeron55 that's the issue, really, if it's thought about as a build system bug 17:14 sfan5 sapier1: outdated? unsafe? 17:14 RealBadAngel they actually felt stopped by you and your vision of it PA 17:14 sfan5 which other do you recommend? 17:14 sfan5 +one 17:14 celeron55 the lack of it allows package maintainers to break it with their flags that they use for making sure dynamic libraries get generated 17:15 celeron55 and yes, also the removal of those like hmmmm said 17:15 sapier1 last libtomcrypt major version is from 2007 last non minor bugfix is 2 years old with respect to development speed in cryptography and security libtomcrypt is useless by now 17:15 RealBadAngel PilzAdam, but it actually turned good, we have more games now 17:15 celeron55 we handle lua just well as a static version, we sure could do that for jthread 17:15 kahrl_ sure, the in-project version can always be static 17:15 PilzAdam RealBadAngel, thats indeed good 17:15 PilzAdam I think we should pack more games with 0.4.8 17:16 sfan5 ^ 17:16 sapier1 I only know of two license compatible maintained crypto libraries ... openssl isn't wanted due to its api ... and I don't know if gnutls api is that much better 17:16 celeron55 kahrl_: no, distributions don't ask about lua 17:16 RealBadAngel technic is not ready for such voyage 17:16 celeron55 kahrl_: because it _works_ without them even knowing about it 17:16 celeron55 they don't go and dig up stuff 17:16 celeron55 they just start looking when it doesn't work 17:16 sfan5 sapier1: what about polarssl? 17:16 RealBadAngel but soon it will be 17:17 sfan5 RealBadAngel: define 'soon' 17:17 RealBadAngel next week i have free 17:17 sfan5 is that after the deadline? 17:17 RealBadAngel weekend is planned to finish shaders 17:17 sapier1 it's gpl ... not sure but seems to be at least more active then the one you selected 17:17 celeron55 kahrl_, hmmmm: so my proposal is: let's make jthread handled exactly like lua is handled 17:17 RealBadAngel next is technic fine tuning 17:18 hmmmm sure 17:18 sfan5 sapier1: celeron55 said libtomcrypt seemed to be most suited 17:18 RealBadAngel we do have already working computer 17:18 RealBadAngel wireless thingies 17:18 celeron55 sfan5: i didn't know about the issues with it 17:18 kahrl_ celeron55: I'm fine with that 17:18 RealBadAngel another texture pack 17:18 RealBadAngel and sounds/ambience 17:18 sfan5 celeron55: what about polarssl? 17:18 celeron55 sfan5: no idea 17:19 RealBadAngel technic_game is already way different to minetest_game 17:20 PilzAdam is anyone against a feature freeze for 0.4.8 now? 17:20 sapier1 polarssl seams to be actively developed but no idea if it can be built for win and posix ... and they do sell a commercial license ... 17:20 RealBadAngel not to mention lstystem trees included. AND moon flower :) 17:20 kahrl_ sfan5: idea: make a mesecons competition "implement SSL in mesecons", then take the best entry 17:21 sfan5 kahrl_: there will surely be many submissions 17:21 sapier1 ok they use dual licensing with gpl2 so we should be fine for license issues 17:21 sfan5 and it will work fast ;) 17:21 RealBadAngel PilzAdam, i could be against, but i left it up to others if they want to get my patches sooner or later 17:22 RealBadAngel so if majority decide to freeze i will too 17:22 sapier1 I don't think anyone of us is capable of reviewing crypographic code ;-) correct me if I'm wrong 17:23 RealBadAngel its too much encrypted for me ;) 17:24 sapier1 but sfan5 I guess cryptography wont be added for 0.4.8 can we delay this discussion? at least if there's need to get 0.4.8 until 14th? 17:24 sfan5 sapier1: k 17:25 sapier1 I don't want to stop you from trying and find concepts that work of course I guess we need multiple tries to get a reasonable version 17:26 hmmmm ahh of course, sapier wants cryptography in 17:26 kahrl_ sapier1: you can't use a gpl library from lgpl code 17:26 RealBadAngel imho we shall merge some stuff during next week 17:26 RealBadAngel and then from 8-14 make the freeze 17:26 sapier1 ok so polarssl is out 17:27 sapier1 gnutls is lgpl ... but I don't know if it can be built on win32 17:27 RealBadAngel so next sunday could be deadline 17:27 RealBadAngel and freeze point 17:28 kahrl_ sapier1: it is, they even provide official win32 builds 17:28 PilzAdam RealBadAngel, I thought you wanted an ultra stable version 17:28 RealBadAngel i do 17:29 RealBadAngel mixed feelings ;) 17:29 PilzAdam and then you think 6 days feature freeze is enough? 17:29 RealBadAngel i will go with the flow 17:29 sapier1 I don't have any Idea kahrl I just looked for recent license compatible crypt libraries ... and I deny it's me who started this discussion this time I only try to prevent adding a useless crypto lib ;-) 17:29 Calinou 6 days feature freeze is enough, minetest isn't that big 17:29 Calinou you get to know immediatly when the game is broken :P 17:30 sapier1 considering feature freeze what about 774,418 and 640 17:30 celeron55 so something called "gnu" is the best for us? lol, wouldn't have guessed 17:30 RealBadAngel Calinou, i not such sure about it. technic game rebuilds meant lately 100k of lines 17:30 kahrl_ Calinou: some obscure parts of the API are only used by a few mods, and they might not get tested immediately with new git commits 17:30 RealBadAngel for a MOD 17:30 sapier1 celeron gnutls is lgpl ;-) even with gnu in it's name ;-) 17:31 celeron55 sapier1: i would have expected GPL, and long beards 17:31 celeron55 the latter would be good though 17:32 sapier1 me too but it isn't it's lglp2.1+ 17:32 RealBadAngel Calinou, https://github.com/RealBadAngel/technic_game/contributors 17:32 RealBadAngel take a look on the numbers 17:32 sapier1 maybe because of openssl's open licensing scheme 17:33 Calinou well, technic is really big, but aren't there a lot of testers? 17:33 RealBadAngel when mods are starting to talk in kilos of lines of code, that means something 17:34 sapier1 I don't have any idea how big mobf is but I don't have any testers ;-) 17:34 RealBadAngel i have managed to attract some decent coders to technic game 17:34 RealBadAngel and im proud of it 17:34 sapier1 but guys I need 418 and 774 for next mobf version (or something similar) 17:34 RealBadAngel theyre doing very good 17:36 RealBadAngel we do have such pearls as programmable computer 17:36 RealBadAngel and lotsa other stuff 17:37 sapier1 :-) interesting ... but I already programm computers I don't want to do it within mt 17:37 RealBadAngel inception, buddy :) 17:37 sapier1 418 automatic entity movment dir facing ... any comments on this? 17:38 sapier1 I know it's most likely only usefull for mobs maybe for vessels too ... 17:39 PilzAdam sapier1, add a comment to protocol version 21 in clienserver.h 17:41 kahrl_ sapier1, shouldn't this check for hypot(m_velocity.Z,m_velocity.Z) > tolerance before changing m_yaw? 17:41 kahrl_ X* 17:42 sapier1 what would be a usefull tolerance? 17:42 kahrl_ dunno 17:42 RealBadAngel PilzAdam, do you think that including technic_game for sake of magazine release would be good? 17:43 PilzAdam I dont know technic_game 17:43 RealBadAngel ofc you do 17:43 sapier1 but there's something else .. i guess it should be normalized 17:43 RealBadAngel its bloated ;) 17:43 kahrl_ normalized? 17:43 sapier1 no no forget about it it's done by atan2 17:43 RealBadAngel 32x texture pack, music, sounds, mesecons, trees, machines and all this stuff 17:44 sapier1 I thought about m_yaw beeing 20 rotations for example 17:44 RealBadAngel and sam II 17:45 RealBadAngel also a computer programmable in forth, force fields, wrench, stained framed glass, concrete posts... 17:46 RealBadAngel ask what technic dont have rather ;) 17:46 sapier1 ok added comment to 418 17:47 RealBadAngel also CNC milling machine is there, circular saw 17:47 sapier1 now the more interesting one 774 17:47 sapier1 I'm not sure if those protocol version changes are right 17:47 RealBadAngel ah, and tetris game :) 17:48 sapier1 basicaly this adds a base speed definition for animations ... telling client what horizontal speed matches the animation at specified framerate 17:48 sapier1 thus client can adjust the animation playback framerate as entities movement accelerates 17:49 celeron55 sapier1: that was already discusses, you have to remove the raised version minimums and test that it doesn't crash with cross-version usage 17:49 celeron55 discussed* 17:49 sapier1 I can't test 20 version combinations celeron 17:50 celeron55 just always keep the latest stable around and you're fine 17:50 sapier1 I'll test against 0.4.7 but I wont test back to 0.4.0 17:50 celeron55 no need to 17:50 celeron55 we don't care about that long network compatibility 17:51 celeron55 (is that written down anywhere?) 17:52 kahrl_ it's quite nice to be able to connect down to protocol version 13 17:53 sapier1 currently it's min version 13 we're at 21 ... so it's roughly 0.4.0 ... I'd need to test for days to be really sure everything beeing allowed works 17:53 sapier1 kahrl yes it's nice but I'm almost sure it hasn't been tested for quite a view versions 17:54 nore I am thinking about texture packs 17:54 nore I could add them with only a few lines of code 17:54 kahrl_ sapier1: there are a few things that don't work, obviously 17:54 nore would it be a good idea? 17:54 sapier1 we already have texture packs nore 17:55 kahrl_ for example some valid node showing up as unknown when you connect to an old server with a new client 17:55 nore no, but textures packs that stay in a folder, with a menu option to say what texture pack to choose 17:56 nore that we don't have 17:57 sapier1 in a folder? 17:57 celeron55 https://gist.github.com/celeron55/6133668 17:57 celeron55 that's a patch that makes jthread behave like lua behaves if luajit isn't available 17:57 proller celeron55, like this ? https://github.com/proller/minetest/compare/ver ! 17:57 sapier1 nore if you put a texture pack within a folder to textures it's in a single folder 17:58 nore sapier1: you put a folder with the TP inside the textures folder, the name of the folder being the name of the TP 17:58 nore and then you can choose in the main menu what folder is to be choosed for images 17:59 proller celeron55, i can push it in 2 commits, for nodebox and disk 17:59 sapier1 ok so you want a texture pack selection option in settings? 17:59 celeron55 proller: i don't see any change 17:59 nore sapier1: yes 18:00 sapier1 good idea I suggest using a dropdown 18:00 celeron55 proller: what did you change since last time? 18:00 nore just need to change tile.cpp 18:00 sapier1 why? 18:00 sapier1 texture pack is set in settings? 18:01 proller celeron55, since _READ and SerializeNetworkSpecific -- nothing 18:01 kahrl_ celeron55: ah, I thought you meant just compiling a static jthread if the system jthread isn't found 18:01 kahrl_ why remove the option to use a system jthread? 18:02 sapier1 because we need more features than system jthreads provides? 18:02 celeron55 like hmmmm said, because jthread is such a small wrapper and nobody benefits at all from a system version 18:02 proller celeron55, oh, find your reply 18:02 RealBadAngel quick thought: easy texture pack selection shall be made before we go magazine 18:02 kahrl_ sapier1: we don't really, I just fixed that 18:02 celeron55 and by doing our changes to jthread itself, we can have code in more logical places 18:02 celeron55 that's hmmmm's view on this issue 18:03 celeron55 i'm quite neutral other than that i want this fixed in way or another 18:03 kahrl_ celeron55: system jthread removes a few seconds of compile time after I do a make clean 18:03 kahrl_ small benefit ;) 18:03 RealBadAngel and magazine pack imho shall consist of a few chosen texture packs 18:04 celeron55 kahrl_: you'll have to fight this with hmmmm unless someone comes up with compelling arguments :P 18:05 celeron55 kahrl_: is there any possible chance that we need to modify jthread's internals that can't be accessed from outside? 18:05 celeron55 same to hmmmm of course 18:05 celeron55 the answer to that could allow a decision to be made 18:06 Jordach RealBadAngel, Vanessa's HDX, Misa's(?), PixelBox 18:06 RealBadAngel and Sphax's one 18:06 Jordach no 18:06 kahrl_ celeron55: I don't claim to know the future so... no comment from me 18:06 RealBadAngel why? 18:06 celeron55 i'm kind of siding hmmmm on this, because always having our own causes less effort for us, because every platform other than windows provides well-working pthreads 18:06 Jordach there is no re-dist allowed 18:06 RealBadAngel ah ok 18:07 RealBadAngel can mine go in? 18:07 celeron55 a conditional build system easily overweights jthread itself :P 18:07 celeron55 -t 18:07 VanessaE don't forget Minetoon and Sphax 18:07 celeron55 but otoh doing it your way is more what i would have originally done 18:07 RealBadAngel sphax is excluded, at least if ashenk ask owner 18:08 RealBadAngel i would like to see it onboard 18:08 celeron55 this is a stupid issue really 18:08 celeron55 we could as well flip a coin 18:08 celeron55 it just doesn't matter 18:08 kahrl_ right :) 18:08 RealBadAngel its most complete and fine texture pack out there 18:08 sapier1 ok I updated 774 tested client046 -> server 04(8) as well as server 046 ->client04(8) 18:09 sapier1 I didn't have a 0.4.7 available but I guess 0.4.6 will do too 18:09 RealBadAngel VanessaE, ofc next to HDX ;) 18:09 sapier1 btw 774 fixes animation blending too 18:09 celeron55 hmmmm: if you merge my patch in yourself, you'll get it your way 18:09 celeron55 hmmmm: we can't choose 18:09 celeron55 8D 18:09 RealBadAngel but as for me HDX is too... girly style :) 18:10 VanessaE :P 18:10 RealBadAngel hahaha 18:10 RealBadAngel you know what i mean :) 18:10 kahrl_ do any other devs have a preference about the jthread issue? 18:10 RealBadAngel i dont 18:11 RealBadAngel not my playground at all 18:11 celeron55 they really have no reason to care at all 18:11 sapier1 imho we should implement threads ourselfs or fork jthread and integrate it completely 18:11 RealBadAngel we do care 18:11 celeron55 sapier1: basically my patch integrates it completely, in all ways that actually matter 18:11 RealBadAngel but we are highly expertized in our areas 18:12 RealBadAngel so celeron55, please dont say we dont care 18:12 sapier1 considering the work this separate lib causes integration may also safe time ;-) 18:12 celeron55 kahrl_: aren't you afraid for even a tiny bit about build system bloat? 8) 18:12 sapier1 save 18:12 celeron55 because of 50 line libraries 18:13 RealBadAngel if you were about to touch shaders now i could argue with you. 18:13 kahrl_ eh, I don't see it often so I don't care about bloat there :) 18:13 hmmmm huh 18:13 celeron55 but when it breaks, it's horrible because nobody is familiar with it 18:13 sapier1 integration also eases feature updates 18:13 RealBadAngel life is life 18:13 RealBadAngel things broke 18:13 hmmmm anyway, we should remove the system JThread option if only for the purpose of removing dependencies 18:14 hmmmm more dependencies = bad 18:14 RealBadAngel we can all live with it 18:14 kahrl_ if you add /usr[/local]/include/jthread to the search paths it should only break very rarely 18:14 sapier1 that's what I meant with integrating 18:14 kahrl_ hmmmm, but from another point of view, more object code duplication = bad 18:15 kahrl_ that's the distro maintainer's view 18:15 celeron55 but it's so little object code 18:15 celeron55 it doesn't matter 18:15 kahrl_ how about this: allow using the system version for now, and integrate it if we need to modify it again for some reason 18:15 hmmmm karhl, sure, except in this case the distro maintainer might want to take a look at JThread in particular 18:15 RealBadAngel PilzAdam, have you reconsidered option for each game having own tiltle music score? 18:15 celeron55 kahrl_: no, nobody remembers that and it gets just messed up 18:15 sapier1 I've seen comercial software with three different thread implementations within a single application ;-P 18:15 hmmmm so my arguments: 18:16 hmmmm 1). it allows us to remove an external dependency 18:16 hmmmm 2). we have total flexibility over the jthread implementation 18:16 VanessaE Jordach: sphax is cc-by-nc-nd, but you're allowed to distribute it. 18:16 hmmmm 3). it gives us an organizational advantage 18:16 Jordach oh 18:16 RealBadAngel VanessaE, so we can include it? 18:16 hmmmm 4). it reduces build complexity 18:16 VanessaE RealBadAngel, Jordach: yes, I just now downloaded and looked at the license declaration. 18:17 RealBadAngel friggin great :)) 18:17 VanessaE (which contains the "human readable text, and a link to http://creativecommons.org/licenses/by-nc-nd/3.0/legalcode ) 18:17 celeron55 there's a name for this issue though 18:17 celeron55 it's the... hmm, i don't remember it 18:17 sapier1 +1 for integration and name change 18:17 RealBadAngel i will add normal and heightmaps for it just 18:18 sapier1 what about mtthread ;-) 18:18 celeron55 this can be compared to someone bringing the plans of a nuclear reactor to the front of you asking whether to build it, and someone building the plans of a bike shed in front of you asking whether to build it; guess which one you are going to argue about? 8) 18:18 celeron55 bringing* 18:19 Jordach i'm going to complain about the bike shed; i know nuclear physics 18:19 RealBadAngel celeron55, have you actually tried to build reactor in mt? :) 18:19 sapier1 I don't need another bike shed lets build the reactor 18:19 * celeron55 looks at RBA racing besides the point in his trollface car 18:19 RealBadAngel :) 18:20 Jordach if the a-team can build a tank out of a taxi, we can build a reactor out of a bike shed 18:20 celeron55 can we decide based on the fact that we want to be evil (obviously we want to), so this allows us to be evil against distro maintainers 18:20 RealBadAngel i just mentioned it because you can build a reactor in mt already :) 18:21 sapier1 no we're not evil we're not google 18:21 celeron55 if there's nothing else to think in the rest of our lifes, we can always silently laugh at them as they didn't know they were shipping countless copies of bundled jthreads 18:21 celeron55 8) 18:22 RealBadAngel celeron55, it looks like youre light years behind of what modders have done to the engine already 18:22 RealBadAngel jordach would call it "fun fact" 18:22 RealBadAngel :P 18:22 sapier1 celeron we can reimplement jthread too if it's so important to be good ;-) 18:23 kahrl_ has nobody mentioned that we should really use boost and boost::thread? 18:23 sapier1 *looking for something to throw at kahrl* 18:23 PilzAdam RealBadAngel, I tend to say: do what the fuck you want to your "magazine version" but dont bother the rest of us with it 18:23 celeron55 boost::videogame 18:24 * RealBadAngel hands sapier1 an empty bucket 18:24 sapier1 empty???? a platic one am I right? 18:24 * Jordach throws ingots at sapier1 18:24 Jordach there. 18:24 kahrl_ tomato ingots? 18:25 sapier1 I hope it's at least copper ;-P 18:25 sapier1 but I'd prefere gold jordach 18:25 RealBadAngel PilzAdam, so i can read that youre against this release? 18:25 Jordach sapier1, i was throwing air ingots 18:25 Jordach to make an air sword 18:25 kahrl_ you hacker you! 18:26 PilzAdam RealBadAngel, no, I just dont care about a random magazine 18:26 celeron55 kahrl_: how about you just merge your thing because it's already tested 18:26 RealBadAngel PilzAdam, its not fuckin random. its the biggest polish one 18:26 sapier1 I do care about it it's a big opportunity if we can use this chance why not do it? 18:26 kahrl_ celeron55: most were against it, so I'm not going to 18:26 PilzAdam re the jthread thing: I agree on what gets merged first 18:27 sapier1 but we should push mod developers to integrate their mods to moddb too 18:27 RealBadAngel i managed to get they approval and you say you pay no shit about it?? 18:27 celeron55 kahrl_: but roughly as many were against mine too 18:27 celeron55 and mine isn't tested 18:27 celeron55 and we have to choose some of them 18:28 sapier1 rba just ignore pilzadam he's strange the last days must be the heat ;-) 18:28 RealBadAngel its fucking sad 18:28 Jordach PilzAdam, New People = New Devs / Modders, New People = Word of Mouth = More people, loop; 18:28 celeron55 RealBadAngel: you shouldn't care about what some dev cares; people have their priorities 18:28 celeron55 RealBadAngel: you obviously care a lot more because you're polish 18:29 kahrl_ we should get a release out soonish regardless of the magazine thing 18:29 Jordach remember, the polish go in their thousands 18:29 RealBadAngel celeron55, you have actutally learned me the most 18:29 sapier1 celeron55 imho we should try do help as much as possible 18:29 RealBadAngel i was about to quit the community 18:29 RealBadAngel because of you 18:29 sapier1 of course if there's no chance to get it stable we shouldn't rush it 18:30 RealBadAngel but then VanessaE helped me realize i shouldnt 18:30 PilzAdam celeron55, MinetestBot has a !rand function 18:30 RealBadAngel i stayed, and im doing what i love to do 18:30 PilzAdam we can use it to decide on the jthread thing 18:30 sapier1 LOL 18:30 kahrl_ design by rng, still better than design by committee :D 18:31 RealBadAngel so, sorry, PA, im immune to your tricks 18:31 celeron55 hmmmm: can you promise to merge either kahrl_'s or my code, because you're the one who made the problem in the first place 18:31 RealBadAngel i was trained by c55 18:31 Jordach who isn't PilzAdam 18:31 Jordach PilzAdam, you should stop with the: I'm elite, GTFO 18:31 RealBadAngel and magazine version is going to be released in all its glory 18:32 RealBadAngel for us all 18:32 sapier1 rba is 14th date of release or last change? 18:32 RealBadAngel 14th is the date for files 18:32 RealBadAngel dvds have to be burnt 18:32 sapier1 ok so we should target our final build at 10th 18:33 Jordach RC1 = July 8th 18:33 Jordach RC2 = July 9th 18:33 Jordach Release = 10th 18:33 RealBadAngel as i was said, if we dont supply files by 14th 18:33 RealBadAngel all will be delayed by month 18:34 RealBadAngel and no excuses 18:34 celeron55 well that's not bad at all, lol 18:34 sapier1 of course not but are there many open issues right now? 18:35 RealBadAngel mc has far more issues 18:35 RealBadAngel without any wishing well 18:35 kahrl_ we shouldn't use MC as an excuse 18:36 RealBadAngel but we shouldnt fear also 18:36 kahrl_ if we did, we wouldn't have a modding api yet :P 18:36 sapier1 are there any other changes that should be added expect of those I already suggested (and weather fixes) 18:36 RealBadAngel we are solving them way faster mc did 18:37 RealBadAngel i catch myself sometimes, sayin whoa! thats a thing mc doesnt have 18:38 Jordach_ we got the modding api first 18:38 RealBadAngel and we are really close to border 18:38 RealBadAngel soon they gonna chase us 18:38 sapier1 but we don't play those feature cards very well 18:38 celeron55 well the racers aren't really even on the same track 18:39 RealBadAngel we miss PR 18:39 RealBadAngel dont you think? 18:39 kahrl_ we don't really need PR as we don't make any money from this 18:39 sapier1 and usability my littl bro still tells me minetest is to complicated 18:39 RealBadAngel we have to sell ourselves 18:39 sapier1 I guess the all new mod download store will draw a lot of ppl to minetest 18:39 RealBadAngel sapier1, exactly 18:39 PilzAdam we don't really need PR as we don't make any money from this <- yep, exactly that 18:40 RealBadAngel thats why i was screamin for installer 18:40 RealBadAngel click n play 18:40 RealBadAngel even for kids 18:40 RealBadAngel because kids know what minecraft is 18:40 PilzAdam the installer is actually more complicated than the .zip 18:40 sapier1 guys everyone doing oss wants to get his work used so don't tell me you don't 18:40 sapier1 lol pilzadam did you ever try it? 18:41 kahrl_ sapier1: not necessarily; some do it just for learning purposes, for example 18:41 celeron55 my view on everything is that diversity is best, but diversity can't be put in the same bucket 18:41 RealBadAngel listen, 5 yrs old girl 18:41 RealBadAngel saw my work screen 18:41 kahrl_ and since they don't plan to make any money from it they release it as oss 18:41 RealBadAngel and said youre playin minecraft? 18:41 sapier1 yes kahrl and forget about it after they got their grades 18:42 RealBadAngel i said no, its minetest 18:42 kahrl_ yeah, and that's not bad 18:42 kahrl_ if others else is interested 18:42 kahrl_ oops 18:42 celeron55 some money flow would stabilize minetest a lot, but otoh the lack of money kind of makes it what it is 18:42 VanessaE RealBadAngel: I got the exact same reaction from a ~10 year old neighbor kid who used to stop by the window to talk 18:42 kahrl_ if others are interested in it they can continue developing it 18:42 RealBadAngel whats the difference,she asked? 18:43 sapier1 kahrl I'm doing this for learning too of course but I do want others to use it 18:43 RealBadAngel we DEFINITELY need dumb easy interface 18:43 sapier1 rba that's what the new mainmenu is targeted at 18:43 kahrl_ sure, I'm not saying noone cares about others using it 18:43 sapier1 who did try mod download store? 18:44 celeron55 RealBadAngel: make an alternative main menu for dummies 18:44 celeron55 and make an installation package that uses it by default 18:44 RealBadAngel celeron55, you mean whole the world outside 18:44 celeron55 then throw that all over the internet (maybe even with a different name) 18:44 VanessaE sapier1: the one in the "Mods" tab? 18:44 sapier1 mainmenu can be switched by configuration option 18:44 RealBadAngel because all the hardcore you already have there 18:44 sapier1 the one showing up as soon as you press "download" yes 18:45 VanessaE sapier1: it looks ok but is totally useless without content :P 18:45 sapier1 exactly :-) but it's easy 18:45 sapier1 and I'm next to sure content will be added fast as soon as this is released in a stable version 18:45 RealBadAngel if im alone can bring here ten players 18:46 RealBadAngel only by makin game easier 18:46 RealBadAngel it means thats a prority 18:46 VanessaE sapier1: extremely important is it needs feedback. when I click "Download", I should immediately see a "Fetching the list of mods, please wait..." 18:46 VanessaE and then the list should appear. 18:46 RealBadAngel and you guys should stop tech talks 18:47 PilzAdam sapier1, the translation thing should be done before we release 0.4.8 18:47 RealBadAngel and make the game friedly 18:47 sapier1 if I remember correctly modlist is already fetched if you press download 18:47 RealBadAngel for any new user 18:47 BlockMen RealBadAngel, just removing https://github.com/minetest/minetest/blob/master/src/environment.cpp#L2245 does not work because it get "reset" in #L2272. But when removing there liquid too you "drown" in air because it is set=true for every node 18:47 PilzAdam RealBadAngel, why dont _you_ do anything about a more user friendly UI? 18:47 RealBadAngel and this shall be a point 18:47 sapier1 if we release as of 10th that's almost impossible 18:47 BlockMen so that must be changed too like i mentioned ;) 18:47 VanessaE sapier1: yes, but it causes the program to hang while it does so - no feedback saying "ok, that button works, but I need time to fulfill your request" 18:48 sapier1 does it? didn't know about that ... I'm gonna have a look at it 18:48 kahrl_ why aren't fetches of that kind done in a background thread? 18:49 kahrl_ or asynchronously (seems more complicated than the thread) 18:49 sapier1 because there isn't a way to do that right now 18:49 PilzAdam RealBadAngel, I mean, you come here several times, say "everyone stop doing what you currently do, and make the UI more user friendly, this is the highest priority"; and what do you do? nothing 18:49 RealBadAngel PilzAdam, i havent changed the logic 18:49 RealBadAngel afaik its your job 18:49 VanessaE kahrl_: I've been asking the same since the Public Server list was added, as it does the same damn thing - and if you lack a 'net connection, the whole game is unresponsive for quite some time when first started. 18:50 RealBadAngel dont fuckin blame me for your faults 18:50 VanessaE s/game/program/ 18:50 VanessaE which makes modding a real pain in the ass if you're offline. 18:50 RealBadAngel i just changed it to use m_breath 18:50 kahrl_ you can change it to use the favorite server list by default 18:50 sapier1 someone needs to decide which one until 10th both are impossible 18:50 kahrl_ but yeah, it ought to be fixed 18:50 RealBadAngel if it is faulty it means you have fucked it up 18:51 PilzAdam RealBadAngel, wut? 18:51 sapier1 either localization or 5 second timeout fix 18:51 RealBadAngel wut wut 18:51 VanessaE kahrl_: I believe the hang happens even if "favorites" is the default focus 18:51 RealBadAngel use english 18:51 kahrl_ it does? I never checked the code and my connection is fast enough to not notice it 18:52 PilzAdam RealBadAngel, you seem to be talking about drowning, why do you talk to me then? 18:52 VanessaE kahrl_: I don't notice it normally - but when I went offline for a few days recently, boy howdy did I notice it :P 18:52 sapier1 almost szre vanaessae as curl uses a timeout to abort if no connection is possible 18:52 RealBadAngel because you pointed it? 18:52 VanessaE but I can't be 100% certain if it was focused on favorites or public at the time, already forgot. 18:52 PilzAdam RealBadAngel, no, BlockMen did 18:53 VanessaE sapier1: that timeout seems to be somewhere north of 30 seconds, which is plain unacceptable since it hangs the program entirely. 18:53 BlockMen I was, whats wrong? 18:53 RealBadAngel PilzAdam, i dont know whats wrong bout you but you alwayas raise my blood pressure 18:54 RealBadAngel anyway, im not responsible for shape of drowning code 18:54 BlockMen RBA, if i said sth wrong, sry, i just wanted inform you (cas i thought it could interest u) 18:54 Jordach_ ladies ladies; hambags down, now get back to compiling and coding 18:55 RealBadAngel i just exposed variables to lua 18:55 * kahrl_ hands RBA some beta blockers 18:55 sapier1 ok ok the timeout is only used for json fetching 18:55 RealBadAngel kahrl_, thx, i already was applied with one 18:55 RealBadAngel one week vacations 18:56 * VanessaE hands RealBadAngel some calcium channel blockers :P 18:56 PilzAdam RealBadAngel, I am still confused why you suddenly start to talk about drowning when we where talking about user friendly UIs 18:56 RealBadAngel PilzAdam, shit, you have linked me to the code 18:57 PilzAdam no, it was BlockMen 18:57 RealBadAngel sorry 18:57 * RealBadAngel is calming down 18:57 RealBadAngel again, im sorry, please forgive me PilzAdam 18:57 PilzAdam you know, BlockMen and me are different persons 18:58 RealBadAngel no shit 18:58 BlockMen we are? 18:58 BlockMen dont say so :D 18:58 RealBadAngel rly? 18:58 sapier1 sorry favorites are a json value too ... can you check value of timeout vanessae? 18:58 Jordach_ i always thought BlockMen == PilzAdam 18:58 BlockMen i know Jordach -.-' 18:58 kahrl BlockMen even fits the voxel concept 18:58 sapier1 minetest.conf curl_timeout it's in ms 18:59 RealBadAngel okay guys 18:59 RealBadAngel lets put those aside 18:59 RealBadAngel i have bigger problem 18:59 VanessaE sapier1: that value is not set in my .conf (and is not explained in the .example, either) 18:59 RealBadAngel each rendered tile needs variables 19:00 BlockMen kahrl, pure accident :P 19:00 RealBadAngel in GLSL shaders such variable is called attribute 19:00 RealBadAngel irrlicht engine doesnt support such variables 19:01 RealBadAngel so if i have a tile, i cannot get modifiers for it easily 19:01 RealBadAngel i figured out how to solve it 19:02 RealBadAngel im creaing textures with irrlicht 19:02 RealBadAngel and setting pixels as data 19:02 Jordach_ hax 19:02 RealBadAngel yup 19:02 kahrl you can get *some* parameters by using a different vertex type 19:02 kahrl attributes* 19:02 RealBadAngel but im sure that proper data reach the shader 19:03 kahrl video::S3DVertexTangents for example 19:03 sapier1 it's set in defaultsettings.cpp ... ppl shouldn't change it but you could try setting it manually 19:03 RealBadAngel kahrl, no 19:03 RealBadAngel theres no way to pass attribute 19:03 RealBadAngel at least i couldnt find a way 19:04 RealBadAngel irrlicht just doest support it 19:04 kahrl let's add irrlicht as a static bundled library that we can modify :D 19:04 RealBadAngel i do have tangent space calculations done in cpp 19:05 RealBadAngel but no way to interface it 19:05 RealBadAngel so i created texture interface 19:05 RealBadAngel since textures are not limited 19:06 RealBadAngel i reserved a channel for settings 19:06 RealBadAngel and one for data 19:06 kahrl textures are limited, that's why some people get messed up inventory textures 19:06 RealBadAngel no they are not 19:07 VanessaE kahrl: no, I believe that's some weird off-by-1 error 19:07 RealBadAngel at least when we talk irrlicht 19:07 kahrl I guess those people won't use fancy shaders anyway 19:07 RealBadAngel shader no matter how fancy is a material type 19:08 kahrl VanessaE: could be, but why does it only happen for a few people and always with weak GPUs? 19:08 RealBadAngel material can have unlimited number of textures 19:08 VanessaE kahrl: I am not sure at all, I've been trying to figure that out forever 19:08 RealBadAngel i think limit is 64k 19:09 RealBadAngel but anyways 19:09 RealBadAngel theres no fuckin way to pass tile dependent data to shader 19:10 RealBadAngel you can pass globals ofc 19:10 RealBadAngel but not tile specific 19:11 RealBadAngel i managed to solve it with setting new texture 19:11 RealBadAngel holding a data which uniform sampler can get 19:11 kahrl if you make a separate texture for each tile you will have an insane number of mesh buffers and an insane number of material switches in the rendering routine 19:12 kahrl sounds like trouble 19:12 RealBadAngel not really 19:12 kahrl it might be faster to use the same secondary texture for multiple tiles and use a secondary texture coord to pass an index into it 19:12 RealBadAngel im makin 10-15bytes long textures 19:13 RealBadAngel which are deleted on nexr render approach 19:13 kahrl I haven't measured it, but I'm always told that material switches are slow 19:13 RealBadAngel hmmm 19:13 RealBadAngel dunno 19:14 celeron55 yes they are slow, and the speed of them variaes by multiple orders of magnitude between different GPUs 19:14 RealBadAngel definitely its a problem 19:14 kahrl e.g. in OpenGL it usually implies a whole bunch of commands to the video card to set the new material options 19:14 RealBadAngel if you guys have better solotution just let me know 19:14 celeron55 eg. i know some old laptop users that got totally crap perforamnce out of MT when the texture atlas once got inefficiently used, while i saw no difference on my old laptop 19:14 kahrl as I said, reuse the same texture and use a secondary coord 19:15 kahrl which is basically the idea of a texture atlas 19:15 celeron55 and in modern rendering you don't even consider whether it might be fast or not, because every command will take forever compared to bulk rendering speed 19:15 RealBadAngel ah 19:16 RealBadAngel kahrl, we are facing another problem now 19:16 RealBadAngel i have already forgot bout atlas 19:16 RealBadAngel what bothers me now is tangent space 19:17 RealBadAngel i need this vector for every plane 19:17 proller celeron55, serialize with net version: https://github.com/proller/minetest/compare/ver , but now finding how to pass version to deserialize... 19:17 RealBadAngel i CAN do this in shader, but its resource heavy 19:18 RealBadAngel and is buggy 19:18 celeron55 proller: just add a version in the data itself like in most places? 19:18 celeron55 if that's an issue otherwise 19:19 proller now finding where to get version on client... 19:19 celeron55 (it probably is, because most of it is designed around versions being put in there) 19:19 celeron55 proller: remember how nodebox did it? 19:19 kahrl RealBadAngel: ah, irrlicht only has one type for vertices with a secondary texture coord and another with tangent vectors 19:19 kahrl but none with both, that sucks 19:19 celeron55 proller: do it in the same way; that's why there are the 1 and 2 versions now there 19:20 celeron55 (better not invent anything new for it but rather rework everything at the same time if so is decided) 19:20 celeron55 oh actually 19:21 celeron55 there *is* a problem 19:21 celeron55 it's that you really shouldn't read even a version byte on client if it's an older version 19:22 * celeron55 facepalms deeply 19:22 celeron55 yeah, that's a terrible job you've got there, try to find a sane way 8) 19:22 celeron55 we'll all pity you after this is merged 19:22 kahrl RealBadAngel: can you encode the tangent vector in the secondary texture? 19:23 celeron55 proller: you can read it actually though, as it's the last thing in the data, just try-catch the serialization error 19:24 proller versions hell 8( 19:24 RealBadAngel kahrl, yes, i did that 19:24 kahrl please, when all is done, someone put up a page on the wiki that explains to dummies like me what has been done to the serialization versions 19:25 celeron55 yeah, we're going to need to tidy this up in a large way sometime 19:25 celeron55 kahrl: it's probably not explainable in human language, but maybe it can be attempted... 19:26 celeron55 by the way, where is is supposed to go there? 19:26 celeron55 the dev wiki seems to be badly organized in that people just link some random pages out of nowhere from there 19:27 RealBadAngel celeron55, https://forum.minetest.net/viewtopic.php?id=6737 19:27 RealBadAngel here 19:27 RealBadAngel for example ;) 19:28 kahrl maybe ChangeLog/Some_title_about_serialization_versions and link to it from ChangeLog 19:28 kahrl unless it's meant to be a more general page 19:28 celeron55 i'll try to make some organization in there 19:29 celeron55 and actually i think the only page that actually has been recently used is the TODO page 19:29 RealBadAngel please do 19:29 celeron55 which isn't linked anywhere in there 8D 19:29 PilzAdam celeron55, the changelog is the most used page, I think 19:30 kahrl the changelog is linked from the download page on minetest.net 19:30 kahrl TODO: link TODO from somewhere 19:32 proller celeron55, client doesnt know net_proto_version 8( 19:32 celeron55 proller: do the same thing as the nodebox you did doe 19:32 celeron55 +s 19:33 proller why not leave as now with serialization version? 19:33 celeron55 it is immensely bad to use a disk serialization version for network 19:33 celeron55 you have to get rid of it, it's the worst thing to do for the long term 19:37 celeron55 kahrl: do you think this should be in changelog, or should it be some separate page? 19:37 kahrl separate page, it would get cluttered otherwise 19:38 proller celeron55, nodebox already have version 1 sended from server 19:38 proller now version 2 19:38 VanessaE another user on my server reports crashes/hangs when opening the inventory (probably https://github.com/minetest/minetest/issues/831) 19:38 celeron55 proller: yes, just include a version similarly in the function you added 19:38 celeron55 start it from 1 19:39 proller but MapBlock doesnt have separate transfer version now 19:39 proller adding new field will broke something 19:39 celeron55 it doesn't have to, as it's the exact same data as is on disk and the server and client agree on what disk serialization version they use for mapblocks 19:40 celeron55 you only break some random development versions network compatibility 19:40 celeron55 that is a non-issue 19:40 kahrl about VanessaE's bug report 19:40 celeron55 you gain compatibility with previous releases which is the goal 19:41 proller will try... 19:41 kahrl are there any threads besides the main thread and the mesh update thread that call ITextureSource::getTexture? 19:41 kahrl TextureSource's request queue doesn't currently identify the caller 19:42 celeron55 i'm going to call this new wiki page "Dev Log" 19:45 nore about texture pack selection, what do you think of this? https://github.com/Novatux/minetest/commit/a8aa531a85ebf10913355fb445e10f0656fc9838 19:49 VanessaE nore: I don't see how that's too helpful without the GUI part to go with it. 19:50 nore I will soon do it, just need to know how to make the listdir of the textures folder, since lua has no listdir command 19:50 sapier1 and texture_path == texture_pack 19:50 sapier1 what should be the difference between texture path and pack? 19:51 nore texture path is where the textures are, texture pack is the name of the texture pack 19:51 VanessaE but, I guess as long as it doesn't interfere with using textures/all as we do now it's okay 19:52 nore no, it does not change textures/all, but textures should not be in texture_path 19:52 sapier1 no no no it's not ok it's adding another way to add a texture pack this is going to be a nightmare to handle backgrounds overlays ... 19:53 nore sapier1: how do you list the worlds, so it is possible to list texture packs? 19:53 sapier1 worlds are scanned by minetest within specific folders 19:53 VanessaE nore: that said, why not just change texture_path to point directly to the desired folder? 19:53 sapier1 no you need to implement your own get_texturepack_list function 19:53 VanessaE handle that with the GUI, use the last directory component as the texture pack name 19:54 nore VanessaE: that could be done, but we would still need a setting for the place of texture packs 19:54 VanessaE e.g. texture_path=/home/user/texture-packs/my_kickass_pack -> "my_kickass_pack" in the GUI 19:55 VanessaE nore: then let the user choose a folder directly with a normal file dialog. 19:55 sapier1 any you can already read a list of directorys in mainmenu 19:55 VanessaE don't overcomplicate it, and don't ask the user to change how they manage their texture packs 19:55 sapier1 use get_dirlist with "/homeuser/texture-packs" for example 19:55 nore is there any builtin file dialog for now? 19:55 proller celeron55, like this? : https://github.com/proller/minetest/compare/ver 19:56 sapier1 menu_lua_api.txt 19:56 sapier1 L152 19:57 VanessaE no no no no no - point to $HOME and let the user navigate to where THEY keep their texture packs 19:57 VanessaE don't just assume they're in some specific folder, they never will be 19:57 sapier1 so you suggest a browe button instead of a dropdown vanessae? 19:58 VanessaE sapier1: precisely. 19:58 sapier1 I thought we wanted to make the gui EASY? 19:58 sapier1 how to explain user what folder to select in file select dialog? 19:58 VanessaE "Browse" -> [file dialog opens, with a clear, large title saying "Select a texture pack folder"] -> user clicks on a folder and hits "OK". 19:58 VanessaE how much easier does it need to be? 19:59 VanessaE don't cater to complete morons 19:59 sapier1 what is a texture pack folder? 19:59 VanessaE if you can't even navigate around your machine with a file manager/file dialog, you need to put your computer awat 19:59 VanessaE away* 19:59 sapier1 no this is only clear to you vanessa ;-) not to the common noob 19:59 VanessaE sapier1: the common noob needs to learn, this is computers 101 stuff 19:59 nore sapier1: how do you use engine.file_open_dialog? 20:00 VanessaE sapier1: if they can find their minetest folder to open and run the game, they can find where they downloaded the texture pack to. 20:00 sapier1 did you ever see win8? navigation isn't expected by current users ;-) 20:00 nore menu_lua_api.txt is not clear 20:00 sapier1 ? .... how more clear can it be? you can specify a formname and a caption? 20:01 VanessaE sapier1: no, never seen it, and I steadfastly refuse to dumb the game down that far (if I had the option anyway) 20:02 sapier1 if you get something except nil user pressed accept 20:02 sapier1 wait ok you're right 20:02 celeron55 kahrl, proller: http://dev.minetest.net/Dev_Log 20:02 sapier1 there's no return value 20:03 celeron55 kahrl: do you think that from here on we should require short write-ups to there of anything that was discussed a lot? 20:03 nore ok, but what is formname for? 20:03 sapier1 forget about the return value and look for a button event 20:03 sapier1 it's like you specify a button 20:03 celeron55 kahrl: (i think so) 20:04 sapier1 with only difference that a file dialog is two buttons at once an formname_accept and a formname_cancel button 20:04 nore sapier: in what file do you use that function? I could understand it more easily 20:05 kahrl celeron55: I wouldn't require it, sometimes the end result of the discussion is clear from the git log 20:05 nore sapier1: ok, I see better what should be done 20:05 kahrl but for difficult technical things that may be important to know later I would require it 20:05 sapier1 modmgr 20:05 sapier1 line 550 20:06 celeron55 kahrl: well, maybe make a topic and link to log if the log is clear? 20:06 celeron55 it'd be good for newcomers to read about all the things... somehow 20:06 celeron55 and all the devs too, as nobody reads this channel all the time 20:07 kahrl ah, yeah that could be done 20:07 nore celeron55: what do you think about texture pack selection: texture packs in a folder, then a dropdown, or folder selection with a file dialog? 20:09 celeron55 nore: a dropdown 20:09 celeron55 it's what we do with worlds too, and mods 20:09 proller celeron55, ver works with old servers 20:09 celeron55 (well, lists, but whatever) 20:09 kahrl this Dev Log is very helpful (even though it still makes my head hurt, haha) 20:10 nore and what would be better: http://irc.minetest.ru/minetest-dev/2013-08-01#i_3231640 20:10 celeron55 proller: what you have now seems quite what i requested; now if we'll get a few others look at it to spot bugs 20:11 proller "Sky" server upped with this, can anybody connect with 0.4.7? 20:12 proller hm.. something wrong with heat&humidity 20:13 proller or not.. 20:14 proller works 20:14 celeron55 lol, i had 0.4.dev-20120606 around... server's PROTOCOL_VERSION is 13...21, client's PROTOCOL_VERSION is 10...10 20:14 celeron55 8D 20:15 celeron55 can't find anything in between 20:15 proller grandpa 20:15 celeron55 i'll build a 0.4.7, this'll take like 5 minutes 20:16 proller somebody plays already except me 20:16 nore should the engine be modified to allow a texture_pack setting: https://github.com/Novatux/minetest/commit/a8aa531a85ebf10913355fb445e10f0656fc9838 20:17 nore or should the gui just change texture_path 20:17 nore ? 20:17 proller spawn catch 0 heat and 0 humidity - very rare point, and its not broken 8) 20:17 proller already -3 20:18 proller why abms on loaded bocks works sometimes with chance=1 ? 20:18 proller its bug or feature? 20:19 nore proller: do you mean they do not always trigger? 20:19 proller always, for examlpe block gets freezed 20:19 proller 16x16x16 20:19 proller its impossible 20:20 proller because chance=20 20:20 proller or 16x16 area gets snow or rain 20:21 kahrl nore: I'd say texture_pack 20:21 nore ah, you mean that sometimes an ABM is executed as if chance=1? 20:21 proller yes! 20:21 nore kahrl: sapier ans VanessaE said the other was better 20:21 celeron55 ha, 0.4.7 built 20:21 nore proller: no idea 20:21 celeron55 took more than 5 minutes 20:22 proller using -jX ? 20:22 nore kahrl: what do you think of the commit? 20:22 nore https://github.com/Novatux/minetest/commit/a8aa531a85ebf10913355fb445e10f0656fc9838 20:22 celeron55 proller: -j2, fastest on this computer 20:22 kahrl nore: 2 things 20:23 kahrl 1. the comment in minetest.conf.example seems a bit unclear, maybe use $ in front of setting names 20:23 kahrl 2. the code in tile.cpp could be optimized a little by just getting the texture_pack setting once 20:23 nore ok, doing it... 20:24 kahrl otherwise it looks good 20:24 celeron55 proller: there's one thing still though, if this works: change the commit of serialization.h from "26: MapBlocks contain heat and humidity. Never written; read the same as 25" to "26: Never written; read the same as 25", as it is completely useless to mention anything about weather in there because nothing about weather was written on disk or anything 20:24 kahrl oh, wait 20:24 kahrl another thing, if texture_pack is "" just look in texture_path as it is done currently 20:25 kahrl to avoid screwing up current users 20:27 celeron55 holy shit that was terrible 20:27 celeron55 fps=1 with no free_move 20:27 celeron55 on icey water 20:27 celeron55 i'll get nightmares out of this 20:27 proller its will melt in 3-5 minutes 20:28 celeron55 oh lol, i have shaders enabled 20:28 Jordach celeron55, i used to play 0.2 with 6 fps 20:28 celeron55 no wonder it ran so badly 20:28 Jordach it was fun 20:28 nore karhl: done now 20:29 nore new commit: https://github.com/Novatux/minetest/commit/ecb764a59fd8b5777b686e5341e2a6640efa9d75 20:37 sapier1 why do you still use texture pack? are you really that eager to fix backgrounds overlays headers and footers again? 20:37 sokomine you can get the experience of 1-3 fps even with high-end hardware if you like - just go to a server that got hit by the item duplication bug and idle a bit in one area 20:37 sokomine it would be great if that bug could be found 20:37 sapier1 just use texture_path and create a reasonable gui for it. imho you don't need to change anything within c++ code to do so 20:39 kahrl sapier1: so you suggest to list the contents of the parent directory of texture_path? 20:39 sapier1 one way to do it ... or specify a reasonable default folder to look for texture packs 20:40 sapier1 at least if dropdown variant is chosen 20:40 nore but then we have a problem if there is no texture_path folder 20:40 proller celeron55, 26 desc changed 20:40 sapier1 that's not a problem if you use a reasonable default 20:40 nore and what should be the default 20:41 nore ? 20:41 VanessaE "reasonable default" is easy: $MINETEST/textures/all 20:41 sapier1 MINETESTDIR/textures? 20:41 celeron55 proller: i'm happy with it now; if you can, squash it to two commits; one for nodebox and one for mapblock 20:41 proller щл 20:41 celeron55 i'd rather have someone else test it too 20:41 proller ok 20:41 nore we still need a listdir command 20:42 celeron55 but as far as i can see, it works 20:42 sapier1 no we don't it's already there 20:42 nore where? 20:42 sapier1 get_dirlist? 20:43 nore ah, ok, found it. 20:43 nore I can now make my code. 20:43 sapier1 maybe you should hide e.g. all but not sure about it 20:44 sapier1 be aware that you're not allowed to list all folders for security reasons 20:44 nore all should be renamed to "Default" in the menu 20:44 kahrl hide the whole selection list if texture_path is set to something other than textures/pack 20:44 kahrl ? 20:45 nore when you say all folders, which ones are you able to list? 20:45 sapier1 good ide this can only be done by ppl modifying minetest.conf manuall thus menu doesn't need to support this 20:45 sapier1 all folders expected to contain minetest data + tmp 20:46 kahrl sapier1: offtopic, but allowing tmp there could be dangerous 20:46 nore and do you think there should be a special image in TPs for screenshot? 20:46 kahrl can lua set $TMPDIR somehow? 20:46 sapier1 I guess it can 20:46 nore and then show it in the selection? 20:46 sapier1 but even without this problem tmp may be dangerous on linux too 20:47 kahrl sapier1: should downloaded mods be unpacked in client/tmp maybe? 20:47 sapier1 we'd need our own mt tmp folder to avoid this 20:47 kahrl err cache/tmp 20:47 kahrl it already exists and it used for sounds 20:47 kahrl is* 20:47 celeron55 haha lol, the ogg loader 20:47 celeron55 it's still there 20:48 sapier1 yes I don't see any reason why this shouldn't be possible 20:48 sapier1 still mainmenu isn't that critical to require instant action 20:49 kahrl when we are bored during feature freeze we could change it 20:49 sapier1 bored? :-) 20:50 sapier1 I try to find a way to allow vanessae's async events in menu right now ... I thought it'd be a 50 liner ... 20:50 kahrl you surely already know there are no bugs in minetest so there's nothing to do during feature freeze ;) 20:51 BlockMen from when until when is feature freeze? 20:52 kahrl BlockMen: not decided yet 20:52 proller celeron55, 2 commits: https://github.com/proller/minetest/compare/ver 20:53 BlockMen kahrl, ok 20:54 nore texture pack selection should be merged before feature freeze, I reckon 20:54 celeron55 i'd be relieved if somebody took a look at that one ^ 20:55 proller celeron55, and something with defaults tune: https://github.com/proller/minetest/commit/5038bc00564913b2db5f1839ad52083c7e08414c 20:56 sapier1 nore depends on when feature freeze will be done and what you're going to change ;-) 20:56 nore only builtin/mainmenu.lua (and perhaps default_settings.cpp) 20:58 celeron55 proller: i don't like it that everyone else is so busy now or went away that they're not looking at these 20:58 celeron55 you shouldn't like it either, if there's anything we've learned 21:00 celeron55 anyway merge the compatibility thing, i've looked at it so long that i'm pretty sure it works well enough 21:00 proller щлфн 21:00 proller okay 21:01 celeron55 while looking at proller's things, i've noticed this time bomb by hmmmm: https://github.com/minetest/minetest/blob/master/src/mapgen.cpp#L694 21:01 celeron55 and the corresponding serialize 21:02 celeron55 if mapnode serialization changes some day, schematic compatibility will silently break because it uses that constant rather than the actual version number 21:02 celeron55 it should be changed to 25, probably 21:03 celeron55 let's not do it now though 21:03 celeron55 but it's wrong anyway 21:04 proller celeron55, but inside deSerializeBulk version is checked 21:05 celeron55 the check is useless 21:05 celeron55 the version input doesn't come from the data 21:05 celeron55 the schematic is versioned separately 21:09 BlockMen has someone an idea what could cause an formspec error in my builds? ingame all labels are missing, but in the new formspec-menu all is shown correct 21:10 proller celeron55, in long term backward compatibility is evil 21:11 celeron55 proller: in on-disk formats, it's simply required 21:11 celeron55 proller: in everything else, too much of it is evil 21:11 proller required one-way convert from old versions 21:11 celeron55 that's compatibility 21:12 celeron55 it's evil, isn't it? 21:12 celeron55 anyway the way we handle most compatibility simply sucks currently 21:13 proller yes 21:13 proller whats about https://github.com/proller/minetest/commit/5038bc00564913b2db5f1839ad52083c7e08414c 21:13 celeron55 we could have the same amount of compatibility with much less trouble with better solutions 21:13 Jordach BlockMen, sounds like sapier1 buggered up formspec again 21:14 BlockMen Jordach, no 21:14 proller but ned to switch everything to key-value 21:14 BlockMen i have talked with him bout that already, and it works in e.g. PA builds 21:14 sapier1 I didn't do anything to ingame formspec except of some sanity checks ... those result in elements not beeing shown at all but not text missing 21:15 celeron55 proller: you can do the cloud thing as long as it looks the same by default, but notice that there are some mgparams stuff in there that you don't probably want 21:16 celeron55 altough, as usual, it would be healthy to wait for others to look at it too 21:16 proller in mgparams only liquid_finite save as you want 21:16 BlockMen maybe wrong ft version 21:16 BlockMen what version you use sapier? 21:17 celeron55 proller: but it isn't used? 21:17 proller cloud with this can be moved upper or lower with saving default visual speed and size, but if upper than 500 - too ugly clipping starts 21:18 proller celeron55, used and works as water_level 21:18 celeron55 how can it work now without it then 21:18 proller with this patch it saved to map meta 21:20 celeron55 uh oh 21:20 celeron55 well if it works, it's good; but those aren't supposed to be in the same commit 21:20 celeron55 they're completely different things 21:20 proller PilzAdam, https://github.com/proller/minetest/commit/5038bc00564913b2db5f1839ad52083c7e08414c -- with increasing liquid update interval, to max 5000 blocks per secont 21:21 proller celeron55, okay, will split it 21:21 PilzAdam how finite liquids a mapgen param? 21:21 PilzAdam +s 21:22 proller its not really mapgen, but it for saving per world for not broking old worlds 21:22 PilzAdam and I dont see why increasing the cloud height is neede 21:22 PilzAdam +d 21:23 proller because for example it can used for rain-snow upper limit 21:23 BlockMen hmm...seems fess has same problems 21:23 BlockMen https://forum.minetest.net/viewtopic.php?id=6769 21:23 celeron55 PilzAdam: changing the cloud height isn't needed, but i don't see how it's bad to allow someone to do that without getting their clouds look odd 21:23 proller and 120 is too low and too easy to reach 21:23 sapier1 VanessaE could you plz check if this prototype results in better favorites list behaviour for you? https://github.com/sapier/minetest/tree/asynchronous_events 21:24 celeron55 oh 21:24 proller i'm for 1000+ clouds, but its now impossible with clipping 21:24 celeron55 that also adjust the defaults 21:24 celeron55 i better go to sleep, i'm not noticing anything 21:24 celeron55 proller: don't change the cloud height without approval of many 21:24 proller ok 21:25 proller PilzAdam, and clouds when you inside it - buggy 21:25 celeron55 it is a fun part of the game to be able to reach clouds easily 21:25 PilzAdam also, liquid_update 0.2 is way too fast for lava 21:25 proller and musch more buggy when you in high mountain, and look at cloud in cave 21:26 proller 0.4 better? 21:26 PilzAdam yea, 0.4 is better 21:27 PilzAdam also increase liquid_loop_max to 10000 again 21:27 PilzAdam a problem with the water update is that there is >1 second lag sometimes, even in singleplayer 21:27 PilzAdam so increase the update interval doesnt really help here 21:28 proller with 10000 no sense to touch upddate 21:28 celeron55 ...so are you making it so that most of minetest's network consumption will be due to... liquids? 21:28 celeron55 there are reasons why it's 1.0 and not 0.00001 21:29 proller maybe limit mapblock update to 1s ? 21:29 celeron55 and what do you gain with that? 8) 21:30 proller faster finite liquid transform 21:30 sapier1 great we're updating fluids faster than mob states :-) 21:31 proller simple mobs now like to jump in lakes 21:31 PilzAdam I kinda get the imprission that Minetest sucks at everything that is dynamic 21:31 sapier1 they always ran into water 21:31 celeron55 PilzAdam: it does, that's no secret 21:31 proller 8( 21:32 sapier1 774 and 418 inprove this without adding (lots of) additional network overhead 21:32 sapier1 :-) 21:34 celeron55 well anyway, minetest kind of can cope with fast changing mapblocks by prioritizing transfers of them 21:35 celeron55 but try to be careful, it's easy to make it just kill itself with eg. 10 players or so 21:35 sapier1 why not make client predict some things? 21:35 celeron55 that's the only way to practically make smooth things 21:36 sapier1 I guess client could calculate fluids almost as precise as server does so updates may not result in big differences as long as latency isn't tooo bad 21:36 proller maybe make dynamic update interval depending on queue size, like if >10000 -- 0.1, >5000 -- 0.3 <1000 -- 1, 21:36 celeron55 well... it's probably hard to interpolate between liquid states in a smooth way 21:37 PilzAdam proller, that wouldnt be good, the liquid "speed" depends on the update interval 21:37 PilzAdam it would give incosistent gameplay 21:37 sapier1 dynamic update interval assumes you're liquids are the only thing that need a lot of cpu performance 21:37 proller PilzAdam, if queue big - speet gets slooow 21:38 proller and if < 1000 - yes, it can affect 21:38 sapier1 still mobs could use as much as possible cpu power to get smarter too ... so you'd need to add prioritys for high cpu tasks 21:39 celeron55 there should be pie charts shown on the client of what uses server CPU and client CPU 8) 21:40 sapier1 yes and preferably mod specific on server ;-) 21:40 PilzAdam there should be more pie too 21:40 celeron55 people would be more interested to keep their code good if they saw how bad it was 21:40 celeron55 now they just throw it in and think "oh minetest is just bad, not my code" 21:41 sapier1 or say mobf is slow because latency shows up for mobs not for other mobs 21:41 sapier1 other mods of course 21:43 proller celeron55, why abms sometimes can run with forced chance=1 ? 21:43 celeron55 like when a block has been unloaded for a long time and gets loaded? 21:43 proller yes 21:44 celeron55 that's a feature 21:44 proller its feature> 21:44 celeron55 a bit hard to control feature though 21:44 proller its possible to disable for some abms? 21:44 proller its makes buggy flooding rain 21:45 celeron55 i think not; that might be reasonable i guess... i also think it's been asked before but not enough for the lazy to get doing it 21:45 sapier1 rain per abm? 21:46 proller https://github.com/proller/minetest-mod-weather/blob/weather/weather/rain.lua 21:49 proller but it hart to catch without fast and flying privs 21:50 proller aabm.chance = chance / intervals 21:55 nore sapier: I finally have a GUI, but textures do not change until minetest is restarted 21:55 nore do you know why? 21:55 nore I even tried to remove the contents of cache/media, but no sucess... 21:57 kahrl proller: mathematically speaking, that formula isn't sound 21:57 kahrl let p = 1/chance, P = 1/aabm.chance, n = intervals 21:58 proller if(aabm.chance == 0)aabm.chance = 1; 21:58 sapier1 in menu they don't change because the path is cached in game they should change as it's read on game startup 21:58 nore no, but texture inside game 21:58 kahrl the probability that at least one out of n random bernoulli events with probability p each succeeds is 1-(1-p)^n 21:58 proller if big intervals ->chance=1 21:58 kahrl that should be 1-P 21:58 sapier1 after exiting game menu should be changed too 21:58 nore you run the game, exit to menu, change TP, re-run game 21:58 kahrl P( 21:58 kahrl P* 21:59 nore textures have not changed 21:59 kahrl right now it calculates P = 1 / ((1/p) / n) = n*p 21:59 proller kahrl, stooop 8) 21:59 sapier1 sorry don't have an instant answer you'll have to investigate 22:00 nore what is that MutexedMap class? 22:00 nore looks like it is used as a cache too 22:00 sapier1 never heared about that one 22:00 kahrl sapier1, nore: is g_texturename_to_path_cache ever cleared? 22:01 nore I don't know, I cleared cache/media, but it looks like that should be cleared (and it is probably the origin of the bug) 22:01 sapier1 I'd expected everything to be cleared after game exits but that's an assumption 22:02 nore cache/media is not cleared 22:02 nore I must go 22:02 nore bbl 22:03 sapier1 yes but it uses hashes to decide if update is required 22:12 hmmmm ack 22:13 hmmmm celeron, good point i guess, the proper way to fix this would be to serialize the bulk serialization version as well and use that 22:13 hmmmm but I don't see anything wrong with simply hardcoding 25 22:14 hmmmm i don't think i was really paying attention when i put that there 22:30 PilzAdam hmmmm, https://github.com/PilzAdam/minetest/commit/4ec5185ed24b6c2c17d77595ee885fe60093a247 22:32 hmmmm aww... 22:32 hmmmm i feared someone would try to add something like this 22:33 hmmmm could you please just add multiple ore definitions instead? 22:33 PilzAdam no 22:33 PilzAdam they have different seeds then 22:33 PilzAdam I want to generate gravel as ore in dirt_with_grass and dirt 22:33 PilzAdam (not in minetest_game) 22:33 hmmmm then how about adding an ore flag to share the same seed or something 22:33 PilzAdam why dont you like the wherein list? 22:34 hmmmm because it's inside of a speed critical loop 22:35 hmmmm and this is such a rare case 22:35 PilzAdam the average case is wherein.size()=1, is it really that slow then? 22:37 hmmmm could you at least fix up the brace style and so on then? 22:37 hmmmm also what's with the return statement on 96 22:38 PilzAdam if the oredef is invalid then wherein is just set to AIR, not need to loop through the others then 22:39 hmmmm oh, that's true 22:42 PilzAdam fixed the brace style: https://github.com/PilzAdam/minetest/commit/32ea9ece297fea8b89288d619fc8ae85d76ff4ac 22:45 hmmmm line 101, the { would be on the previous line 22:45 hmmmm line 164, you have all the operators clunked together 22:45 hmmmm if(lua_istable( on line luaapi.cpp:684 22:45 hmmmm other than that it's good 22:47 PilzAdam whats wrong with line 684? 22:48 hmmmm no space after 'if' 22:49 PilzAdam is it actually part of our code style guidelines to put a space there? 22:49 hmmmm yes... 22:49 hmmmm if isn't like some sort of function, it's a control statement 22:50 hmmmm the reason why you don't see me constantly yelling about that is because that particular style is present in code everywhere else, but not really new code (like mapgen or emergemanager or so on) 22:50 hmmmm so, you could use that style if it fits in with what's already there, but in general don't 22:50 hmmmm like i said, follow the linux kernel code style 23:01 proller astyle --style=attach --indent=tab --keep-one-line-blocks --keep-one-line-statements --indent-classes --pad-oper *.cpp *.h 23:02 proller kill all human! 23:03 VanessaE Who do you think you are? Bender? :) 23:04 PilzAdam hmmmm, is a "biomes" field in oredef planned? 23:24 hmmmm pilzadam, actually, yes 23:40 BlockMen ok, i have rly no idea why the ingame formspecs have no label in my builds (and fess's too). i have tried different versions of gettext, freetype and irrlicht 23:40 BlockMen always same result... 23:40 BlockMen so if anyone has an idea, let me know plz 23:40 PilzAdam since when do fess's builds have this problem? 23:41 BlockMen https://forum.minetest.net/viewtopic.php?id=6769