Time Nick Message 08:25 nerzhul pgimeno the camera offset problem you mension means that we just badly coded and we must refacto in order to have a single function to calculate 10:46 sfan5 wut 10:49 sfan5 imagine coming here with a shitty attitude that you know everything better and then not even sticking around to receive an answer 10:49 sfan5 all while ignoring the explanation that was given to you on the bugtracker 10:55 MTDiscord who? 10:55 sfan5 https://irc.minetest.net/minetest-dev/2021-02-12#i_5784467 10:59 Krock not everybody wants to read... 11:56 nerzhul ouch 12:13 pgimeno $ egrep -Rl --include='*.cpp' --include='*.h' camera_offset | wc -l 12:13 pgimeno 18 12:14 sfan5 that's not too much 12:24 pgimeno enough to cause a rebase headache 13:48 pgimeno does BS affect Lua code at any point? 13:49 Krock only in the attachment distance 13:49 Krock I think there it's exposed directly 13:50 sfan5 though exposing it there was a mistake 13:51 pgimeno how bad would it be to break compatibility of attachment distance? 13:51 sfan5 why do you need to break it? 13:51 Krock pretty bad for advtrains I'd say 13:51 pgimeno I'd like to get rid of BS at the same time as camera_offset 13:52 sfan5 well the api code can just divide the numbers it gets by 10 13:52 pgimeno that's what I was thinking, though it sounds kind of artificial 13:53 sfan5 why get rid of that anyway? 13:54 pgimeno because it means meaningless multiplications and having to take something unnecessary into account while coding 13:58 pgimeno meaningless multiplications *and* divisions, the latter being typically slower 14:01 sfan5 yeah but I'm wondering 14:01 sfan5 what was the original reason to add it and why is it no longer relevant 14:06 Megaf Hi all 14:06 pgimeno https://github.com/minetest/minetest/blob/master/src/constants.h#L69 14:06 Megaf sfan5, so, why you don't want to get make install fixed? 14:07 sfan5 it is working as designed, with RUN_IN_PLACE=0 the folder structure is different because it needs to go into e.g. /usr 14:08 sfan5 pgimeno: hmm so looks like it's only there so people don't forget to use the helper functions 14:09 pgimeno I don't see that as a compelling argument 14:09 rubenwardy it was definitely an accident 14:10 pgimeno rubenwardy: what was? 14:10 rubenwardy the 10* for attachment 14:10 pgimeno ah 14:10 rubenwardy would you could do is add `attach` rather than `set_attach` which uses node space 14:10 rubenwardy `set_attach` is bad grammar 14:10 rubenwardy should be `set_attachment` or `attach` :) 14:10 rubenwardy this could probably just end up confusing 14:11 rubenwardy if Minetest had a versioned api, then you could do this much cleaner 14:11 rubenwardy versioned API meaning local minetest = get_api(39) 14:11 pgimeno best would be to use a fixed 10.0 value for the old set_attach regardless of BS, and deprecate it 14:11 sfan5 soon: set_attach, set_attach2, attach, real_attach 14:11 sfan5 ;) 14:12 rubenwardy yeah, that would be the idea with adding `attach` - to not break `set_attach` 14:13 pgimeno please be very careful when doing that, you don't want to repeat the case of http://wiki.secondlife.com/wiki/LlXorBase64Strings http://wiki.secondlife.com/wiki/LlXorBase64StringsCorrect and http://wiki.secondlife.com/wiki/LlXorBase64 14:13 pgimeno (the latter of which is still buggy btw) 14:17 rubenwardy updated #10943 14:17 ShadowBot https://github.com/minetest/minetest/issues/10943 -- Fix short_description fallback order by rubenwardy 14:18 rubenwardy If there's a definite bug with a feature, you can fix it in the next release. However, you will need to be careful with the design itself 14:21 rubenwardy Should support for `alpha` be readded given that the engine never had a deprecation warning for it? It could use `^[opacity` on the tiles if the `alpha` field shouldn't be changed 14:24 sfan5 does the limited support it has now break any games/mods/etc.? 14:24 rubenwardy it broke nodecor 14:24 rubenwardy not sure about other mods 14:24 rubenwardy it feels like quite a messy situation for Warr 14:26 rubenwardy Will older clients connecting to old servers or new servers use the alpha from the texture correctly as `blend`? Your spreadsheet says "val", not sure what that is 14:26 Megaf >" it is working as designed, with RUN_IN_PLACE=0 the folder structure is different because it needs to go into e.g. /usr" I beg to differ, make install should install all libs and folders to wherever INSTALL_PREFIX is defined, the default is indeed /usr/ or /opt depending on the software. INSTALL_PREFIX overwrites that. 14:26 Megaf am I the only one annoyed by a broken make install? 14:27 Megaf Or even the only one who uses make install 14:27 sfan5 distros literally make their packages with "make install" since 8 years 14:27 rubenwardy cmake . -DCMAKE_INSTALL_PREFIX=/usr/local 14:27 rubenwardy this is a cmake thing 14:27 rubenwardy not sure if it reads INSTALL_PREFIX 14:28 Megaf https://cmake.org/cmake/help/latest/variable/CMAKE_INSTALL_PREFIX.html 14:28 sfan5 the point is by using RUN_IN_PLACE=0 you are saying you want the folder structure for a system-wide install, which is >>not<< the same structure as for a local installation 14:28 Megaf rubenwardy, yep, I peant DCMAKE_INSTALL_PREFIX 14:28 rubenwardy ok 14:30 Megaf I got into building my own stuff now, and in this Minetest pause of mine I got into FlightGear, and they use cmake too, and for its depencied too, so there I use CMAKE_ISNTALL_PREFIX for OpenSceneGraph, PLIB, SimGear and FlightGear, cmake even correctly find the libs required where I pointed with the prefix. 14:31 Megaf using cmake install prefix makes run in place redundant, the end behaviour is the very same 14:31 Megaf with the advantage of creating a clean install without having unecessary git and source files in the final binary destination 14:32 rubenwardy the install prefix is typically used when making .deb/.zip packages for this 14:32 rubenwardy you do want run in place, because you're expecting the conf and games to be in the same folder 14:32 rubenwardy otherwise, they go to ~/.minetest 14:33 Megaf good point 14:33 sfan5 (you can actually use CMAKE_INSTALL_PREFIX and RUN_IN_PLACE=0 but I don't think that's what is wanted here) 14:34 Megaf With all other software and libraries I tested what happened is what I wanted, to have the required built lib and binaries in a directory specified by me 14:35 sfan5 It is what happens with Minetest, too, but in your issue you claim that "everything is missing" is missing despite the 'bin' and 'share' directories being installed as designed 14:36 sfan5 s/" is missing/"/ 14:37 Megaf That is the bug, only bin and share were copied, other essential bit waren't 14:37 sfan5 is this a joke? 14:37 rubenwardy share contains games and other stuff 14:37 sfan5 bin and share contain all those "essential bits" 14:38 sfan5 it broke nodecor < link? 14:39 rubenwardy https://irc.minetest.net/minetest-dev/2021-02-07#i_5782654 14:39 Megaf If I am mistaken then I do deeply apologize, I will double check that here right now. 14:40 sfan5 please do since that is what I was trying to communicate the entire time 14:40 sfan5 rubenwardy: oh since you wrote "nodecor" (missing e) I thought that was something different 14:40 rubenwardy oh right lol 14:41 sfan5 but did this break nodecore in a way that water suddenly turned opaque? 14:41 sfan5 because that is the special function of the 'alpha' property 14:41 rubenwardy I believe that was the issue - they had alpha set to something like 72, with opaque textures 14:43 sfan5 I briefly tested that actually when developing but couldn't remember if water was supposed to be opaque so I didn't do anything about it 14:55 Megaf Ok, I figured out what the problem was here, I wasnt using RUN_IN_PLACE=1 with CMAKE_INSTALL_PREFIX 14:55 Megaf CMAKE_INSTALL_PREFIX + RUN_IN_PLACE=1 results in 14:55 Megaf $ ls Minetest 14:55 Megaf bin builtin cache client clientmods debug.txt doc fonts games locale minetest.conf.example mods textures unix 14:55 Megaf Which is precisely what I wanted 15:01 sfan5 RUN_IN_PLACE=0 would work too fyi 15:28 pgimeno what C++ version are we at, again? C++11? 15:28 Krock ^ 15:30 pgimeno is it OK if I add that to https://dev.minetest.net/Code_style_guidelines ? 15:31 sfan5 sure 15:33 pgimeno done, please don't forget to update it when the standard requirement is raised 15:51 pgimeno when will the version be bumped? when Stretch goes EOL? 15:58 Krock when there's a need for it 16:15 Krock sfan5: would you mind pushing "The Fix" (#10949) ? 16:15 ShadowBot https://github.com/minetest/minetest/issues/10949 -- Segfault with malformed node texture string 16:16 sfan5 I'll do that sooner or later 16:16 Krock the second "return" needs "c" too btw 16:17 Krock great thanks 20:58 sfan5 required reading about the attachment mess https://github.com/minetest/minetest/pull/10928#issuecomment-778676120 21:04 Krock wow thanks for this research. Good to know that the server is supposed to have the authority in first place. I'll update the PR tomorrow 21:05 Krock or rather open a new PR since most changes in 10928 are due to SmoothTranslator integration 21:07 sfan5 sounds good