Time Nick Message 00:05 kilbith actually the first slot is 1.2 in width until it decreases to 1.0 on the last slot of the line 00:08 kilbith this is `size=1.1,1.1` for example: https://i.imgur.com/hDsPaXv.png 00:36 numzero on textures with transparent pixels: just use premultiplied alpha (both for rendering and mipmap generation) 00:38 numzero that would eliminate most artifacts (as well as the need of clean_transparent_textures) 00:41 sfan5 if the engine does mipmap downscaling itself we could also just fix that 00:41 sfan5 don't think it does though (?) 00:41 numzero fixing that means multiplying by alpha before averaging 00:42 numzero and dividing by sum of alphas 00:42 numzero only to multiply again during rendering 00:43 numzero also if one texture may be used both with and without alpha, mipmaps need to be different to look correctly 00:44 sfan5 hm 00:44 sfan5 I suppose there's a toggle in OpenGL for premultiplied alpha? or change the shader for that? 00:44 numzero almost 00:45 numzero different blending mode is needed 00:45 numzero I hope Irrlicht allows using the one we need 00:45 numzero and premultiplying alpha on texture generation ofc 00:46 numzero of well, shader change is also needed if it needs to affect transparency 00:46 numzero that’s not a lot of changes actually 00:48 numzero the interesting thing is making different textures for transparent and opaque usages 00:53 numzero I think appending ^[noalpha when required (e.g. when used on an allfaces_optional→solid node) is a good way 00:54 MTDiscord Nba for now, it costs mtg only 11KB 10:04 sfan5 https://github.com/minetest/minetest.github.io/commit/132bc7c0d967fc0eb80cdb047858a24592fb1a46 10:04 sfan5 it has not? 10:04 sfan5 last time I added a feature I even made sure to not use a windows 10-only API 10:04 sfan5 in any case the downloads page should still list 8 and 8.1 10:55 rubenwardy I could have sworn this was a thing 10:58 Krock IIRC we only discussed Windows XP, and perhaps 7 too 11:02 rubenwardy 8 definitely should not have been removed 11:03 sfan5 technically 8 is EOL already but not like it makes a difference support-wise 11:03 sfan5 I already added 8 and 8.1 back to the downloads page 11:05 rubenwardy If it's at all unclear then it should be discussed anyway 11:06 rubenwardy I don't think 7 should be supported, but I'm not the one supporting Windows (I try to avoid that). So really it's up to you two 11:06 rubenwardy Sorry for not checking before doing this 11:10 sfan5 I'd say keep 7 support for as long as practical 11:10 sfan5 ..which will probably be a long time 11:40 nerzhul_ i really think we hsould not support windows xp & vista. What is our current % of 7 users sfan5 ? 11:46 Krock https://kitsunemimi.pw/tmp/serverlist_stats_2020-05-10.txt 11:47 Krock ^ nerzhul_ 11:47 nerzhul_ 8 months old 11:47 nerzhul_ refresh required :D 11:48 Krock 15%, but tendency towards 5% if the decline is linear to 2019 11:48 nerzhul_ 15% of all windows, which means 1% of the population 11:50 Krock Minetest must be really popular on mobile, but there are surprising less issues 11:52 Krock @e​xe_virus Release date is not next Saturday. that's when feature freeze will start. the release happens a few weeks after 11:52 rubenwardy Android and macOS are much higher maintenance burdens than 7 tbh 12:04 sfan5 nerzhul_: did some quick calculations, it's 16.1% 12:08 sfan5 w10 is 74.8%, 8/8.1 is 5.1% 12:11 sfan5 a bit more actually but the rough picture is correct 12:16 sfan5 by the way, judging from the version strings 5.3.0 can still work on Windows XP 13:16 giov4[m] Hey, today I used the `register_on_chat_message` callback for the first time, and I realized that once a mod handles the message (returning `true`) the other callbacks are no longer executed; but if I wanted to have 2 mods, one that, for example, implements proximity chat and another that implements the prefixes, how could I make it? 13:17 rubenwardy Prefixes can be done using format_chat_message 13:17 rubenwardy Also, modding questions should be in #minetest 13:18 giov4[m] oh right, sorry ?, switching to #minetest 14:06 rubenwardy small PR: #10864 14:06 ShadowBot https://github.com/minetest/minetest/issues/10864 -- ContentDB: Order installed content first by rubenwardy 14:29 rubenwardy merging #10808 in 10 14:29 ShadowBot https://github.com/minetest/minetest/issues/10808 -- Improve irr_ptr by numberZero 15:25 rubenwardy #6765 is long overdue 15:25 ShadowBot https://github.com/minetest/minetest/issues/6765 -- Add more neighbors on mesh update by numberZero 15:25 rubenwardy I'm not familiar with Minetest's lighting code 15:54 numzero (continuing from #minetest) 15:54 numzero nephele: so do you have readelf? 15:55 nephele I haven't yet, i'll checkout the broken commit in a bit and rebuild 15:57 numzero do you have objdump then? 15:58 nephele again, i need to rebuild it for the broken state, and then i can check 15:58 nephele :) 15:59 nephele i have 491: 0000000000000000 0 NOTYPE GLOBAL DEFAULT UND _ZN9PcgRandom5rangeEii for both sky.cpp and noise.cpp 16:00 nephele and a second symbol in sky.cpp that doesn't appear in the noise.cpp list: 492: 0000000000000000 0 NOTYPE GLOBAL DEFAULT UND _ZN9PcgRandomC1Eyy 16:00 numzero heh 16:00 numzero which _ZN9PcgRandomC appear in noise.cpp? 16:00 nephele 117: 0000000000000000 46 FUNC GLOBAL DEFAULT 2 _ZN9PcgRandomC1Emm and 118: 0000000000000000 46 FUNC GLOBAL DEFAULT 2 _ZN9PcgRandomC2Emm 16:01 numzero m is unsigned long and y is unsigned long long, I suppose 16:01 nephele I can also just put the grepped output to termbin if you like :) 16:02 numzero that’s not needed anymore, I think 16:05 numzero https://sourceforge.net/p/irrlicht/bugs/433/ 16:06 numzero client/sky.h includes before including "irrlichttypes.h" 16:07 numzero nephele: try putting the irrlichttypes include first in client/sky.h 16:09 nephele Yeah that seems to work 16:10 nephele I'll try it on master 16:20 nephele yeah, works on master too https://termbin.com/l9z8 16:20 nephele thanks 16:22 rubenwardy lol, the weblate form says "Please contact us in English, otherwise we might be unable to process your request." 16:33 Krock I'd say a vast majority of projects uses English strings as base to be translated, thus it's not so ironic 17:03 MTDiscord Is "not so ironic" a good thing? ? 17:25 nephele numzero: if you want to make a specific pull request for that issue to merge into master i'll happily test it too btw 17:25 nephele just won't make one myself, lacking a gh account :) 17:28 rubenwardy I've updated the change log https://dev.minetest.net/Changelog#5.3.0_.E2.86.92_5.4.0 17:30 nephele huh, interesting, it sais support for bump mapping and such was removed, which seems to be the case, but the menu options still seem to be there 17:32 rubenwardy menu options are gone for me 17:33 nephele i'm on origin/head, didn't use run in place though 17:34 nephele Add support for Haiku OS should probably just be add support for Haiku :) 17:36 rubenwardy > didn't use run in place though 17:36 rubenwardy this will be why 17:36 rubenwardy builtin needs to be up to date 17:45 nephele okay, yeah, with run in place the bumpmapping option is gone 17:49 sfan5 "This was broken on many platforms. " could be more like "These features had fundamental issues, several bugs and were broken on some platforms." 17:49 sfan5 (regarding changelog) 17:49 rubenwardy updated thanks 19:23 kilbith yet another bug with list styling: `style_type[list;size=0.01;spacing=0]` + hovering causes an exception 19:24 kilbith it should be required to all PR authors to test their features in all f** conditions 19:25 kilbith it's been 3 easily reproducible bugs already 19:26 Krock to be fair, since you were so eager about this feature, you could've tested those cases too to help 19:28 kilbith I avoid git apply'ing PR diff 19:28 kilbith too annoying 19:28 rubenwardy I use git hub to create branches for PRs 19:29 rubenwardy it's a bit lazy and tied to GH, but works very well 19:30 nephele kilbith: reallly...? 19:30 nephele it's just a git pull 19:32 rubenwardy you have to find out the branch as well, it's a faff 19:32 nephele that's how git pull usually works, you give it a branch and upstream ^_^ 19:33 kilbith #10866 19:33 ShadowBot https://github.com/minetest/minetest/issues/10866 -- List styling exception 19:33 rubenwardy `git checkout https://github.com/minetest/minetest/pull/10864` is much easier than `git checkout https://github.com/rubenwardy/minetest cdb_installed_order` 19:33 kilbith I realized you can literally paint in the formspec with list styling 19:34 kilbith slot being a "pixel" 19:34 Krock oh no 19:34 nephele rubenwardy: yeah but it does something different 19:36 Krock checkout -b and push is about the same effort 19:39 nephele kilbith: that gives me http://bash.org/?62682 vibes.... 19:39 kilbith ? 19:42 kilbith don't you find that the project is healthier without paramat now? 19:42 kilbith it should be proudly mentioned in the changelog that paramat is out and that's a great feature 19:42 nephele that's a bit rude 19:44 kilbith he didn't even say a word for RBA when he died 19:44 sfan5 ? 19:44 kilbith so today I'm celebrating his quit 19:45 nephele i'm sure that putting slander in the changelog will signal to potential new contributers that minetest is an open and welcoming community 19:46 Krock btw thanks rubenwardy for the changelog so far 19:47 sfan5 did I miss something? 19:47 Krock in the previous ones I usually skipped trivial/non-functional changes bit more details aren't wrong either. that'll give it a better impression for 5.4.0. more updates = better 19:48 Krock !seen paramat 19:48 ShadowBot Krock: I saw paramat in #minetest-dev 9 weeks, 2 days, 3 hours, 46 minutes, and 56 seconds ago saying "merging #10639" 19:48 rubenwardy I've been cutting down the number of mentions, feel free to delete further 19:49 Krock okay sure 19:56 kilbith ouch, the engine badly handles 10e6 micro-slots 19:57 kilbith re-implementing Paint is not for today yet 19:58 nephele not before a drawing api... :) 21:54 lhofhansl Assuming eventually we want to go to zstd compression... Should we keep my change: #10715? 21:54 ShadowBot https://github.com/minetest/minetest/issues/10715 -- Allow configuring block disk and net compression. Change default disk level. by lhofhansl 21:55 lhofhansl Those compression levels are zlib specific and do not translate nicely to zstd. 21:57 lhofhansl And what happened with paramat? He quit? Did I miss something? 22:17 rubenwardy Oh god, I missed those messages. Please keep things civil 22:20 numzero fetching a PR: git fetch origin refs/pull/10864/head # or refs/pull/10864/merge 22:20 numzero ^ kilbith, did you know that? 23:03 kilbith rubenwardy, so did you find a way to remove all unused headers with your IDE? 23:06 kilbith rubenwardy, https://www.jetbrains.com/help/clion/unused-include-directive.html 23:07 kilbith probably 500-1000 LOC less in the whole project 23:07 rubenwardy It removed them but then I got compile errors, I suspect some files rely on other files including things 23:08 rubenwardy And time to sleep 23:21 MTDiscord Wait Paramat quit?!? Bummer honestly. It's always good to have devs, even ones you disagree with 23:47 Zughy[m]1 nobody confirmed it, it sounds more like a rumour. Maybe he's just busy