Time Nick Message 01:35 paramat game#2271 01:35 ShadowBot https://github.com/minetest/minetest_game/issues/2271 -- Add large cactus seedling by paramat 05:18 Unarelith wow just tried #6165, looks so nice, what was the issue with it before the author stopped working on it? 05:18 ShadowBot https://github.com/minetest/minetest/issues/6165 -- Time-of-day-dependent face shading, specular lighting by kaadmy 05:24 sofar not as important as continuing the patch ;) 05:30 * VanessaE sighs 05:40 VanessaE Unarelith: in short, perfect became the enemy of good. 05:40 VanessaE again. 05:51 Unarelith damn, if someone could at least rebase it could be great 05:51 Unarelith but giving the amount of work I don't know if anyone will actually do it someday 07:40 Unarelith did it 07:40 Unarelith but unfortunately it doesn't work with current engine version 07:45 ANAND Wow, thanks 07:46 VanessaE someone should rebase/rewrite that one, and if necessary, implement the shadows one that's linked in the commentsd. 07:46 VanessaE -d 07:49 ANAND Kilbith's PR? 07:52 Unarelith ANAND, don't thank me yet, the PR still doesn't work :'( 07:53 VanessaE ANAND: yeah (#6839) 07:53 ShadowBot https://github.com/minetest/minetest/issues/6839 -- [WIP] Dynamic Shadows by kilbith 07:53 Unarelith VanessaE, iirc this PR wasn't ready at all 07:54 VanessaE doesn't matter. "not ready" is better than "not implemented at all" 07:54 VanessaE in terms of work required to make it mergable. 07:56 Unarelith well, before even considering 6839 I think having a working #7987 would be awesome 07:56 ShadowBot https://github.com/minetest/minetest/issues/7987 -- Time-of-day-dependent face shading, specular lighting [WIP] (#6165 revived) by Quent42340 07:57 VanessaE indeed. 07:58 VanessaE ANAND: you are Quent*? 07:59 Unarelith no, it's me :') 08:00 VanessaE oh ok :) 08:00 VanessaE man, I swear I can't read today 08:02 Unarelith this bug looks so weird 08:02 Unarelith https://i.goopics.net/vvN1Z.png 08:03 VanessaE what's the bug? 08:03 Unarelith it's like when you start making a voxel engine, and you forget to add proper lighting 08:03 VanessaE ohh heh 08:03 VanessaE somehow I didn't notice the lack of shading :) 08:03 Unarelith normal lighting to compare: https://i.goopics.net/55Q3g.png 08:04 VanessaE yeah 08:04 VanessaE I saw the shadow in the cave first, my eyes just kinda skipped the [lack of shading] on verticals 08:08 * VanessaE mentally moves the above misplaced ] :P 08:19 ANAND Lol, VanessaE. I'm ClobberXD :) 08:33 VanessaE heh ok. 09:23 Unarelith ok now I understand why the side faces are bright, but still not close to the real problem hmmm 10:17 Unarelith :D got something, finally 10:19 Unarelith ok wtf I don't get node normals 10:25 Krock oh great. this stupid entity handling code will make us stuck at version 1 forever 10:25 Krock *adds another version variable to the bottom* 10:26 Unarelith Krock, has something involving node normals happened in the past year? x) 10:27 Krock maybe 10:27 Krock much has changed 10:28 Krock nerzhul: do you think unknown entities from 5.0.0-dev worlds would annoy 0.4.x players? 10:29 Unarelith this is #6165: https://i.goopics.net/kK92Z.png (I used `color.rgb = gl_Normal`) 10:29 ShadowBot https://github.com/minetest/minetest/issues/6165 -- Time-of-day-dependent face shading, specular lighting by kaadmy 10:29 Unarelith same thing with #7987: https://i.goopics.net/gNVqD.png 10:29 ShadowBot https://github.com/minetest/minetest/issues/7987 -- Time-of-day-dependent face shading, specular lighting [WIP] (#6165 revived) by Quent42340 10:29 Unarelith the code is the same I checked many times 10:55 Krock Open PR No100. #7989 10:55 ShadowBot https://github.com/minetest/minetest/issues/7989 -- Fix entity rotation in existing worlds by SmallJoker 11:18 p_gimeno Krock: why do you say 'version' doesn't allow incrementally extending the parameter list? isn't that exactly what happened in 0 -> 1? 11:20 Unarelith #7987 finally working :D 11:20 ShadowBot https://github.com/minetest/minetest/issues/7987 -- Time-of-day-dependent face shading, specular lighting (#6165 revived) by Quent42340 11:22 nerzhul can you provide the before/after pictures on the PR itself pleaes ? 11:31 Unarelith nerzhul, pictures are on the original PR 11:31 Unarelith I'm making your changes atm, I'll add some screens later 11:32 nerzhul yes on the final pr can be nice 11:37 Krock p_gimeno: yes, that was probably the change from the old per-sector map to the new map as we know it today 11:37 Krock if you look at the code, then you can see that the case for version >= 2 is not covered. all entities will become unknown on old clients 11:38 p_gimeno Krock: yeah, isn't 5.0 supposed to be forward incompatible? 11:39 p_gimeno as in, no support for moving worlds back to older clients 11:40 Krock for the map it's supposed to be compatible. the per-modpack enabled mods PR has exactly the same issue. 0.4.x clients must support it as well 11:40 p_gimeno but 0.4.x don't support the serialization format anyway 11:40 Krock or at least behave very closely to what's been before 11:40 p_gimeno what am I missing? 11:40 Krock which serialization? 11:40 Krock it does read the fields which it knows 11:40 p_gimeno protocol version as stored in the database 11:41 Krock no, the protocol version is only used for network communication. around that there are various version numbers for each serialization type 11:41 Krock maps have to be protocol-independent 11:41 p_gimeno hmm, let me look 11:42 p_gimeno so you're saying that if I take a 5.0 world and place it in a 0.4 server, it should work, right? 11:42 Krock yes 11:42 Krock it MUST WORK 11:43 Krock although, there may be some unknown nodes, but that's normal 11:43 Krock already happened in previous versions as well 11:44 Krock or did we already break that by the content/lighting changes? I hope not 11:47 p_gimeno ok, yes it does, which surprises me because (1) I didn't think forward compatibility was intended and (2) I've seen that kind of breakage often in the 0.4 series 11:49 p_gimeno so, there's basically no hope of fixing a shortcoming in the serialization format, right? 11:49 Krock I think there should be at least one 5.x release (5.0) which is compatible with the old map format, to give the users the chance to play with the new network and old network code, using the same world 11:50 Krock forward-incompatible map changes are not great but unavoidable 11:51 Unarelith nerzhul, made your changes + added screenshots 11:52 Krock however, the map version itself would have to increase so that there's no data loss when saving (rotation is not so important in this case) 11:53 p_gimeno the version2 hack is a bit worrying though 11:58 Krock p_gimeno: yes sure it is. the other deal would be to turn all objects to unknown for 0.4.x clients on 5.x worlds 11:58 Krock ..or to increase the map version directly which breaks the compatibility entirely 11:59 Krock ^ this would allow to change the serialization format (read old and new, save only new) 12:00 p_gimeno Krock: could you maybe turn the field into a length instead? just in case it stays that way for long, and there's a need to extend it in future to accommodate other changes 12:01 Krock there's still is.good() to see whether the end of the stream was reached 12:02 Krock it would also be possible to use that one instead of the version (although is a bit less secure) 12:04 * Krock wonders why yet nobody increased the formspec version number 14:03 rubenwardy Krock: because it's pointless 14:04 rubenwardy With formspecs, the client version matters - not the server one 14:04 rubenwardy So the server increasing the version is just wrong behaviour 14:04 rubenwardy What if you have an older server and a new client and mod? 14:04 Krock why? that's because mods run server-side and make use of the new formspec elements 14:05 Krock the formspec version will tell older clients that new arguments for the elements can be expected 14:06 Krock the existing element arguments/parameters may not change, though. 14:06 nerzhul Unarelith ty 16:03 rubenwardy Krock: the server doesn't parse formspecs at all 16:03 rubenwardy and mods cannot set the formspec version 16:06 p_gimeno IIUC this is another forward compatibility issue: if you port a new world to an old server, the old clients may have issues with the formspecs 16:07 p_gimeno port ... to -> use ... in 16:07 rubenwardy yeah 16:07 rubenwardy because the world will contain newer formspecs in meta 16:07 rubenwardy which, btw, should be in the node definition instead 16:13 p_gimeno is forward compatibility a goal for 5.0.0? 16:30 p_gimeno how do I compile with clang? do I just change CMAKE_CXX_COMPILER and CMAKE_C_COMPILER, and that's all? or is there anything else to do? 16:31 celeron55 you should be able to just set the CC and CXX environment variables 16:31 celeron55 then cmake tries to act accordingly 16:32 celeron55 (i've never tested this) 16:35 rubenwardy I have 16:36 rubenwardy it's as simple as using env variables, you may need to do a make clean and rm -rf CMakeCache.txt CMakeFiles/ 16:41 p_gimeno thanks 16:42 nerzhul p_gimeno just that is sufficient 16:42 nerzhul cmake .. -DCMAKE_CXX_COMPILER=/usr/bin/clang++ -DCMAKE_C_COMPILER=/usr/bin/clang 16:42 nerzhul and then it works 16:42 nerzhul i use that way since ages :) 16:43 p_gimeno thanks nerzhul, that allows me to not have to look into configuring the build again 16:44 p_gimeno build fails for me in clang 16:45 p_gimeno https://paste.scratchbook.ch/view/3b3a639c 17:02 p_gimeno this is what I have found about that: https://stackoverflow.com/questions/33809441/default-initialization-of-const-qualified-type-with-no-user-provided-constructor 17:03 p_gimeno since we only support C++11, that should be fixed 18:43 Unarelith I get this when I compile in Debug mode: 18:43 Unarelith cc1plus: warning: -Wabi won't warn about anything [-Wabi] 18:43 Unarelith cc1plus: note: -Wabi warns about differences from the most up-to-date ABI, which is also used by default 18:43 Unarelith cc1plus: note: use e.g. -Wabi=11 to warn about changes from GCC 7 18:44 nerzhul yeah i got it too with GCC 8 18:45 nerzhul but if you set ABI to 11 travis fail on GCC 5 & 6 18:45 nerzhul due to unk abi=11.. 18:46 Unarelith uh ok ._. 19:05 Krock will push #7992 in 5 minutes 19:05 ShadowBot https://github.com/minetest/minetest/issues/7992 -- clang build broken in Debian stretch (clang 3.8.1-24) 19:18 Krock pushing after 13 minutes... 19:27 p_gimeno thanks 19:31 Krock yw 20:15 Unarelith nice, all my last revived PRs are working now :) 20:21 Unarelith I really hope you'll have the time to review them for 5.0.0 since they're only adding optional features + a very useful one (leftclick drag)