Time Nick Message 00:00 pgimeno as for UTF-32 not being characters... well, there isn't an immediate relationship between UTF-32 length and displayed width, but certainly there is an immediate relationship between UTF-32 length and string length in codepoints. Rendering Unicode is a different matter, and that's why things like Pango exist. 00:01 pgimeno hecks: I'm looking at the Blender page. Do Blender models appear mirrored with respect to their Minetest counterparts? 00:02 hecks If your coordinate space is correct, they should not. 00:03 hecks You need to export as Z forward Y up X right 00:06 pgimeno if the handedness of the coordinate systems is different, it should 00:06 hecks nothing should appear mirrored if the exporter converts the coordinate space correctly 00:07 pgimeno does it? 00:07 hecks my character's right hand in blender is still her right hand in minetest 00:08 hecks if this is not the case, you messed up your settings or the exporter is buggy 00:08 pgimeno maybe the page should specify that the converter should take care of this conversion 00:08 hecks it does, in the coordinate space section 00:08 hecks always export as lefthanded y-up 00:09 pgimeno I'm asking because I didn't understand that from the page 00:10 hecks if you just follow the instructions (model facing Y+ in blender, export as left-Y) it should all work 00:11 pgimeno my suggestion is to change the phrasing of this sentence: "Be sure to export all your models in Lefthanded Y-up to avoid surprises." to: "Be sure to select Lefthanded Y-up in the exporter to avoid surprises." 00:11 pgimeno I can do it if you agree 00:12 hecks I guess, although I don't know how the B3D exporter calls this space, I don't use it 00:13 pgimeno as phrased, it may be interpreted that it's your job to ensure that you're using the right coordinate system while modelling, and that's where I had trouble understanding and why I asked if you had to have your models mirrored 00:13 hecks no, just set the export options properly 00:13 hecks the original article had some weird advice about making the model face +x, i hope that's a mistake and not some weird b3d quirk 00:14 hecks but a quick look at sam says it was a mistake 00:14 hecks he's facing +y 00:14 hecks he's also modeled in BS space which is disgusting 00:15 hecks right, shit 00:16 pgimeno ok, rephrased 00:17 MTDiscord i didn't model it btw, just textured 00:17 pgimeno BS is BS, the name says it all :) 00:18 hecks how could we eliminate the need to use BS space while modeling entities? 00:19 hecks currently you either have to model in 10x scale or inflate the visual_size 00:19 pgimeno if it's a requirement for compatibility, we can't 00:20 pgimeno there are a number of things that depend on BS=10, and that includes physics too 00:20 hecks added a coordinate scale section 00:20 pgimeno I'd like to get rid of everything BS related, maybe keep a COMPAT_MULTIPLIER applied to some stuff that requires it and that's all 00:21 hecks the stupid thing is that node meshes do not need this scale 00:21 hecks meaning if you want to use a mesh as both a node and an entity, whoops 00:21 hecks you need visual_size = { x=10, y=10 } 00:21 hecks and if that wasn't enough 00:22 hecks say your entity is a skinned mesh - now if you want to give them a hat... 00:22 hecks as a bone attachment that is 00:22 hecks this hat can NOT be scaled 10x, it must actually be 1x 00:22 hecks and then if you ever detach it, it'll shrink back to 0.1x and you'll have to inflate it again 00:23 hecks it's a sordid mess and that's why i'm asking about some sort of deprecation path for it 00:30 pgimeno hmm... new attachment API? 00:30 hecks that wouldn't solve a thing, think about it 00:30 hecks the attachment issue happens because you are forced to apply a scale to the first model 00:30 hecks you either model both in 10x scale or deal with this crap 00:31 pgimeno well, accompanied by some sort of declaration that a certain mesh is "new kind" 00:31 pgimeno where "new kind" means not scaled 10x 00:31 hecks in that case you don't need new attachment API, the flag would be enough 00:32 hecks it would have to scale the actual model instead of the entity (big difference) 00:32 hecks metric_scale = true; 00:34 pgimeno right, so if a new attachment API is not necessary, such a flag for meshes (or a new drawtype) would suffice to solve that particular issue, right? 00:34 hecks just a flag for CAOs 00:35 hecks it would basically mean (after BS space is killed) that the BS emulation should not be applied to this CAO 00:35 pgimeno yes 00:36 hecks I don't think it'll be that difficult to destroy BS, I'll look into it sometime 00:37 pgimeno yeah, my plan was, 1) create a COMPAT_MULTIPLIER, 2) change BS to 1, 3) replace BS in everything that breaks with COMPAT_MULTIPLIER 00:37 pgimeno 4) delete BS 00:40 pgimeno but getting anything done without being in github takes a lot of patience and I don't have enough 00:41 pgimeno also some luck that someone is willing to forward a PR with my code 00:42 hecks I would just delete BS everywhere except I/O because that's the only place it matters 00:42 hecks it only has a place in serialization and being emulated in script API, due to how floats work it's not useful anywhere else 07:05 nrz websockets are tcp based and slow 07:05 nrz i already show 4 years ago that a full switch to TCP just make client moves laggy 07:06 nrz the ordering of the client moves are not adaptred for MMO games 08:45 sfan5 not like you have an alternative if you want to port MT to the browser 09:49 MTDiscord how much left is there to remove from irrlicht now after #48? 09:49 ShadowBot https://github.com/minetest/minetest/issues/48 -- Install icon and .desktop file on Unix systems by jnumm 09:49 MTDiscord that's... not the right repo lol 09:54 MTDiscord @__sub maybe it should be irrlicht#48 ? 09:54 MTDiscord yes 09:54 MTDiscord irrlicht#48 09:54 MTDiscord looks like it hasn't been implemented yet 09:58 sfan5 irr#48 09:58 ShadowBot https://github.com/minetest/irrlicht/issues/48 -- Delete unused features by hecktest 11:03 nrz sfan5: i don't see the point to port MT to browser, it's as huge as rewrite the whole client in rust :D 11:09 rubenwardy webassembly exists. It's probably as huge as adding Android support 11:09 rubenwardy finicky build system and weird graphics 11:12 nrz and redo the network layer to add compat on all servers ? and having nasty bugs/performance issue due to the fact anyone has not a up to date browser ? : 11:12 rubenwardy you wouldn't change the network protocol at all, you'd use a websocket wrapper 11:12 rubenwardy and yeah - it won't be the best experience, I'm against it 11:29 sfan5 we already have "nasty bugs/performance issues" on android 12:11 rubenwardy yeah true 12:31 MTDiscord TCP is no laggier than UDP. In my experience there is a very narrow band of network conditions under which TCP vs. UDP makes a difference. In better conditions they both work fine, in worse, neither has enough to work with. 12:31 MTDiscord I certainly wouldn't switch to websockets for everyone, but as a wrapper for web players it should be fine. 13:00 sfan5 not really 13:00 sfan5 UDP is exactly what you want for games because TCP stalling everything when one packet gets lost is not acceptable 13:01 sfan5 (interactive games that is, you won't need it for chess) 13:02 MTDiscord but that one packet may likely be important, e.g. a node getting placed/dug 13:04 sfan5 it's more likely not to be important, you send e.g. 30 movement packets a second and if one gets lost you just wait for the next 13:09 Calinou people are essentially remaking HTTP on UDP these days with QUIC 13:09 Calinou TCP is OpenGL whereas UDP is Vulkan :) 13:10 Calinou for the web, you can get unreliable transfers with WebRTC but it's a pain to set up compared to WebSockets 13:52 MTDiscord My point about TCP vs UDP is that the differences in practice are much smaller than you would think. UDP's advantage over TCP is really only when you have a network that just arbitrarily loses or reorders packets, but is otherwise fairly healhty which I just don't see much. Usually it seems like people are at the extremes. You have players on fiber who would get perfectly fine performance out of either. You have players on 13:52 MTDiscord satellite for whom the experience is largely ruined either way. There's only a narrow band where the difference is actually important. 13:52 hecks this "narrow band" is cell networks which are very common 13:53 MTDiscord Having used cell networks myself, I can say that that narrow band is PRESENT in cell networks, but cell networks themselves really cover the full range. 13:54 MTDiscord It's also highly dependent on the game. I think the engine should offer options for players without assuming that those options would be intolerable to them without knowing the game design. 13:55 hecks i'm just saying that my internet can drop quite a few packets every now and then 13:55 MTDiscord I certainly wouldn't play CTF with a 150ms ping time, but NodeCore I've tested with ping times in the 1000+ range. 13:56 MTDiscord I would not make TCP or TCP-based protocols/wrappers the default, but I don't want to dismiss them as an option. There are a lot of places where streams can go that datagrams can't. 13:56 MTDiscord It MAY be something exernal to the project, like a proxy, but we MAY want to do it internally if e.g. we want to support something like play-on-web or something. 13:57 sfan5 I'm not against a potential web client, but the advantages are UDP are real and not just in some "narrow cases" 13:58 MTDiscord There's a small community that seems to be growing of users hosting servers on the yggdrasil overlay network because they don't have enough control over their own NAT at the ISP level. That network actually tunnels everything over TCP/TLS. 13:59 MTDiscord Anyway, anecdotally I've never seen enough advantage to UDP that would suggest I should go out of my way and build my own reliability layer to use it over TCP, and I've never seen anything more than anecdotal or "it stands to reason" evidence to refute it. I would be more convinced it's worth real effort if I saw some real-world hard data. 14:00 MTDiscord We would face the same decision deciding between websockets and webRTC for a potential web client. 14:01 sfan5 does webrtc even work for this? I know it does in theory but the primary usecase of it is something else 14:02 MTDiscord It's an interesting question. As far as I understand it though WebRTC is sort of a generic transport and it's not necessarily specific to media? I mean there's no reason why we couldn't lie to it and just say "oh yeah, this is totally a VOIP voice call, it's just a codec you're not familiar with, don't worry about it :-)" 14:02 MTDiscord I think it's mainly for client-to-client communication with the server acting as an introduction point, but there's no reason why an MT server can't be both a server and an endpoint, either. 14:03 hecks irrlicht/#48 is ready, do we merge it? 14:03 ShadowBot https://github.com/minetest/minetest/issues/48 -- Install icon and .desktop file on Unix systems by jnumm 14:03 MTDiscord I think I poked around with it actually, just can' remember what the specific problem I was trying to solve was, and it seemed to have a lot of flex to it. 14:03 hecks bad bot 14:03 sfan5 https://github.com/kyren/webrtc-unreliable 14:03 MTDiscord irr#48 14:03 ShadowBot https://github.com/minetest/irrlicht/issues/48 -- Delete unused features by hecktest 14:04 MTDiscord Hmm, webrtc-unreliable looks pretty interesting. If there were an existing something we could adapt to use in MT then that would make me lean more seriously into the idea of using it... 14:04 sfan5 do we have any mods that use BMP? 14:05 sfan5 I'm somewhat inclined to keep it since it's just so simple 14:05 hecks bmp is supported 14:05 hecks i had to keep it for the null driver to work 14:05 sfan5 "BMP is also enabled because the null video driver would not compile without it." oh missed that 14:06 sfan5 software renderer still exists right? 14:06 hecks it does not 14:06 hecks some remnants of the software renderer still exist for CImage support 14:06 hecks like image blitting and scaling 14:07 sfan5 ah so that's why SoftwareDriver2_compile_config.h still exists 14:07 hecks yup, that's a dep of CBlit 14:08 MTDiscord how much is there left that can be removed from irrlicht now? 14:08 hecks quite a bit but i think i got all the low hanging fruit 14:08 hecks so i'm making a cutoff point here 14:09 hecks any further reductions will involve paying a little more attention when deleting code 14:09 hecks this was a simple grep-delete-repeat 14:09 sfan5 I'll go and test though I don't expect bugs 14:10 sfan5 one more thing though >Also for merging later it'd be good if you could reauthor the commits so that Github can automatically match them to an account. 14:10 hecks how do i do that 14:10 sfan5 or did you want to squash it? 14:10 sfan5 hmm good question 14:10 hecks squash it if you want 14:10 sfan5 no need to bother then 14:10 hecks i added a comment in one commit about not squashing it but i don't think there's any danger there now 14:10 hecks and if it breaks i know what to restore 14:11 hecks it was when i deleted support for loading texture names parsed out of models 14:12 hecks it turns out irrlicht is a relatively well organized pigsty after all, if this only took three days 14:12 hecks it also seems to compile faster than minetest in general, we should be using incomplete class pointers more 14:13 sfan5 pointers are bad in modern c++11 14:13 sfan5 s/11$// 14:14 hecks well, just use incomplete types is what i mean 14:14 sfan5 might also be because irrlicht uses -fno-rtti and -fno-execptions 14:14 hecks more often than not headers do not need other headers 14:15 hecks the thing is that rebuilding a simple change in irrlicht somehow doesn't trigger a cascade where half the objects need updating 14:15 hecks while it's often the case in minetest 14:15 hecks and i'm pretty sure it's thanks to incomplete declarations and possibly the widespread use of interfaces 14:17 sfan5 okay stuff still works 14:17 sfan5 merging irr#48 in 5m 14:17 ShadowBot https://github.com/minetest/irrlicht/issues/48 -- Delete unused features by hecktest 14:18 sfan5 the coolguy_opt.x mesh shows pretty tiny but I guess it doesn't matter 14:19 hecks it's metric, not BS 14:21 sfan5 wait actually let me fix that since it's so easy 14:22 hecks just don't fix it by scaling the mesh node =] 14:22 sfan5 I won't 14:25 MTDiscord should the outdated doxygen windows binary in scripts/doc/ be deleted? 14:25 hecks not unless it's broken 14:26 hecks as long as it works and produces the docs, who cares 14:26 sfan5 there's more that could be thrown away, for example the entire tools folder 14:26 sfan5 but it doesn't really mattert 14:26 sfan5 -t 14:26 hecks curiously there's a copy of the B3D exporter in the tools folder 14:27 hecks if it's outdated, then maybe it should be killed to avoid confusion 15:47 hecks hmmmmm i'm getting some crashes online 15:52 hecks it's not armor, it's not mobs... 16:03 MTDiscord is it maybe tga images? 16:05 MTDiscord TGA is an outdated format that is better served by PNG 16:05 MTDiscord mcl2 uses it 16:05 MTDiscord and? 16:05 MTDiscord nobody cares about mcl 16:05 hecks we don't even officially support TGA 16:05 hecks lua_api.txt only mentiones png 16:05 MTDiscord also that 16:05 MTDiscord true lol 16:05 hecks so this would be a case of relying on undocumented behavior... 16:05 MTDiscord TGA was useful back in the day since PNG wasn't around 16:06 MTDiscord but PNG supports the exact same things and is easier to write libraries 16:06 hecks i'll gdb this particular problem because a lot of servers seem to crash 16:06 hecks something is telling me it's not a file format but some obscure SceneNode behavior i didn't account for 16:06 hecks CTF crashes instantly for example 16:07 MTDiscord just take your time 16:07 MTDiscord however: -216,514 16:07 MTDiscord nice 16:08 hecks oh ok, i'm just getting a segfault in a completely unrelated place before i even log in 16:08 hecks thanks gdb 16:09 hecks yeah wtf, i can't debug like this 16:09 hecks at /usr/lib/gcc/x86_64-w64-mingw32/9.2.0/include/c++/bits/stl_tree.h:798 16:14 hecks can somebody with a less retarded debugger test this 16:15 sfan5 >Reason: We aren't accepting new players right now. Please try again another day or contact us on the forums 16:15 sfan5 I would 16:16 hecks that's on ctf? 16:16 sfan5 it's what ctf.rubenwardy.com:30001 tells me 16:16 hecks try the "Nyanachism" server, it crashes too (after a few seconds rather than instantly) 16:17 sfan5 tried a different ctf server and here https://0x0.st/-VFf.txt 16:17 sfan5 as one might guess this is the case: m_spritenode = (irr::scene::IBillboardSceneNode *) 0x0 16:17 rubenwardy CTF uses sprite entities 16:18 hecks i didn't touch these 16:19 hecks but okay, this doesn't look hard to fix 16:19 sfan5 _IRR_COMPILE_WITH_BILLBOARD_SCENENODE_ is missing 16:19 hecks hm? oh 16:19 hecks i may have swept it up while cleaning up the config 16:20 hecks i'll commit a fix in a minute 16:25 hecks wow wtf 16:25 hecks irrlicht uses crlf... 16:25 sfan5 it does yes... 16:26 rubenwardy also svn and sourceforge 16:36 pgimeno please normalize those 16:37 pgimeno https://stackoverflow.com/questions/15641259/how-to-normalize-working-tree-line-endings-in-git 16:41 pgimeno a repo should always use LF as line ending internally for text files, and convert on checkout 16:43 sfan5 does that generate a big commit that removes and readds each line? 16:44 MTDiscord yeah it would 16:44 hecks i say defer the renormalization until we start absorbing the remains 16:46 sfan5 yes 19:10 rubenwardy I wonder if it's worth doing a much much simpler main menu redesign, without waiting for a new UI framework 19:11 rubenwardy there's multiple issues to solve with a redesign - the look and feels, and the fact that it doesn't promote games/customisation especially well 19:11 hecks could be, there are new formspec features to take advantage of 19:14 celeron55_ well why not 19:14 hecks any design in mind or does that need doing too 19:15 hecks actually the tech doesn't matter that much 19:17 hecks a good layout should work both with forms and with anything that would come after 19:18 hecks so: games and customization as the main focus, anything else? 19:18 rubenwardy I had a mainmenu design based on celeron55's mockups. It would start with a game selection screen. From there, you can either select a game, install a game from CDB, open settings, or join a server. When selecting a game, it would take you to a game-specific menu with singleplayer/multiplayer/content/etc 19:18 rubenwardy The problem I found was that people were confused about the game vs engine difference 19:18 rubenwardy you could join a server using Games > Online or Games > Minetest Game > Online for example 19:18 hecks i would propose something less drastic, a reorganization of the scheme we have 19:19 hecks where the tabs of the main menu and the game list are merged into one 19:19 hecks and there is a landing page 19:20 hecks all represented by icons 19:20 hecks the game list portion of it becomes scrollable once it cannot fit 19:20 hecks keeping it horizontal would give you more space 19:24 rubenwardy I'm not quite sure what you mean 19:24 rubenwardy here's my mock ups (warning, ugly - consider them to be a wireframe rather than an actual example) https://github.com/minetest/minetest/issues/6733#issuecomment-621970748 19:24 rubenwardy but yeah - by "worth doing a much much simpler main menu redesign" I meant compared to this or the alternatives 19:25 rubenwardy I think just showing a choose game dialog as the first thing would improve visibility 19:27 hecks https://a.uguu.se/PGcElo0IVGMt_layout.png 19:27 hecks where landing page contains: a grid of recently used or installed games, labeled links to other tabs, version info, other links like website 19:28 rubenwardy hmm, I can see that working 19:29 rubenwardy yeah that sounds good. I suppose the default game formspec would be the Start Game tab? 19:29 hecks the game formspec has a default but a game can have a menu/form.lua to redefine it 19:29 hecks yup, that would be it 19:29 hecks BUT 19:29 hecks i would also like to suggest servers finally providing a game ID string to the master server 19:29 rubenwardy they already do 19:29 rubenwardy but we don't display it 19:29 hecks that's great actually 19:30 rubenwardy In my redesign, I wanted the server list in the game-specific menus to only show servers matching that game 19:30 rubenwardy it's problematic for Minetest Game 19:30 rubenwardy as so many are heavily modded 19:30 hecks this is exactly what I was about to suggest 19:30 rubenwardy but it's nice for eg: Mineclone2 / NodeCore 19:31 MTDiscord i think there are 6 nodecore servers currently 19:31 hecks a button that takes you to a filtered and customized server browser 19:31 hecks and then separately that "full server list" button that shows you all 19:32 rubenwardy would that button take you to the full server list tab with a filter applied? 19:32 rubenwardy or would it replace the game tab's content? 19:32 rubenwardy or a dialog? 19:32 hecks it could be either really 19:32 rubenwardy I think the former is the simplest in terms of UX, probably 19:32 hecks it could just take you to the full list and apply a filter, and this would be easy to implement 19:32 rubenwardy former=full list tab with filter 19:32 hecks but it wouldn't be impossible to let the game handle it too 19:33 hecks same thing for content installation, though there is a snag here 19:33 hecks do mods yet have a way to specify which game they're meant for? 19:34 rubenwardy not yet 19:34 rubenwardy the ContentDB installer allows you to specify a base game 19:34 hecks that is a bummer 19:34 hecks well uh wait 19:34 hecks so the CDB entries do have it? 19:34 rubenwardy not yet 19:34 rubenwardy it's planned 19:34 hecks because that's the only important bit; you could have a similar thing for content installation that would take you to a filtered view of the content DB 19:34 hecks from the game page 19:35 rubenwardy this is what happens when installing something from CDB with the wrong base game: https://rwdy.uk/5fNO1.png 19:35 rubenwardy yeah, I planned to add that too as it makes sense 19:36 hecks also a somewhat tangential question 19:36 hecks if an mt installation is already able to manage its content by downloading and deleting mod and texture packages 19:37 hecks wouldn't it be nice to have helpers for creating or forking new mods and games? 19:37 rubenwardy as in, a command to make a mod template? 19:38 hecks yes, a button to do that from cdb 19:38 hecks or well, the content menu or whatever 19:38 rubenwardy perhaps 19:38 rubenwardy however, it's not like we have a Java API which needs loads of bootstrap to build using gradle 19:38 rubenwardy and boilerplate to integrate with the API 19:38 hecks no, but i think the forking part would be convenient 19:39 hecks to make a quick and dirty patch to a mod without soiling your usual install of it 19:39 MTDiscord Forking would best be done on github/gitlab though. It seems complicated to have CDB involved in that process... 19:39 hecks i only mean in the local sense 19:39 hecks but uh, i don't know where i'm going with this actually 19:40 hecks only that mt could technically have a built in project manager for mod development i guess 19:40 rubenwardy I'm not sure what that would do tbh 19:40 MTDiscord If players want to hack up their local copy of a mod ... it might be useful to have MT know the difference between an altered copy and an original, for purposes of updates. 19:40 MTDiscord Beyond that, adding too many developer tools is kind of dangerous scope creep for MT. 19:41 hecks back to this layout thing then 19:45 hecks oh right actually 19:45 hecks i think this can just be fullscreen now 19:45 hecks and any logo can be inside the big panel rather than above it 19:53 hecks behold https://a.uguu.se/lKcCBZw6lVyN_landingpage.png 19:57 MTDiscord looks decent actually 19:57 rubenwardy moving the icon tabs to the bottom like this would give a lot more space for games: https://user-images.githubusercontent.com/2122943/83007914-47084f00-a00c-11ea-9fb0-ba0b0c2a5ceb.png 19:57 rubenwardy but yeah, I like the idea of a rich landing page 19:57 hecks it would but..... that's basically what we have now 19:57 hecks and 19:58 hecks the idea is to explode the "single player" tab into individual game icons 19:58 MTDiscord having it so far out of the way feels both out of place and doesnt make it very easy to notice 19:58 MTDiscord I'd change "get more shit at content db" to "get more crap at content db" so that it can be used in an elementary school setting, but other than that, ship it! 19:58 rubenwardy I'm referring just to the icon bar at the bottom there, not the window contents 19:59 rubenwardy but yeah, you're right that it puts some distance which isn't idea 19:59 MTDiscord The icon bar is what im referring to as well 19:59 hecks well this kinda is similar except i placed it on the top, i guess bottom is fine 19:59 rubenwardy I gathered, Green 19:59 hecks there's little difference but uh i guess it gives the logo some breathing room 20:00 MTDiscord I'm a bit curious about how the icon bar would be ordered in the event someone has a buttload of games installed. MRU maybe? 20:00 MTDiscord shouldnt the game have control over where stuff goes 20:00 rubenwardy unfortunately not with this design. They'd control their own page at most 20:00 rubenwardy maybe they'd hack the global layout if the code isn't sandboxed 20:00 MTDiscord I'm okay with opening up main menu control to games gradually. We don't want to make them responsible for it all right away either. As long as we're moving in the right direction. 20:01 MTDiscord thats fair 20:01 MTDiscord choose something and go with it 20:01 MTDiscord get it made and merged 20:01 hecks it's fine for a game to only control the contents of its panel 20:01 MTDiscord It'd be better to get something small and mergeable and start to build momentum quickly instead of designing something too good too soon. 20:01 hecks it should not touch the main bar 20:01 hecks oh and 20:01 rubenwardy the top icons bar is likely to be the first version, as we don't have the ability to pin stuff to the edges yet 20:01 hecks contrary to that example posted i'd give more prominence to the global server browser 20:01 MTDiscord I was mostly referring to the title, since the title is property of the game 20:01 MTDiscord Frankly if you can just make MTG not bundled and have a sane way to get to Content and pick your first game, then I'm already pretty happy :-) 20:03 sfan5 "the game formspec has a default but a game can have a menu/form.lua to redefine it" I suggest postposting this since the mainmenu code isn't sandboxed so we can't "just" do this 20:03 sfan5 oh ruben already mentioned 20:03 hecks perfection https://a.uguu.se/3J0OWkCaMVnd_final.png 20:03 sfan5 a menu redesign would be much worth it even without this 20:03 rubenwardy "_final" 20:03 hecks =] 20:03 hecks i don't see how this could be improved further honestly 20:04 sfan5 lens flare 20:04 MTDiscord more comic sans 20:04 rubenwardy we need more blue 20:04 rubenwardy err 20:04 rubenwardy blur 20:04 rubenwardy but blue is good too 20:04 MTDiscord blur only the blue parts 20:05 MTDiscord bluer? blure 20:05 v-rob Why are there two settings buttons? 20:05 v-rob Just in case you get lost? 20:05 rubenwardy my design aesthetic is just rounded corners and blurs https://renewedtab.com/ 20:05 sfan5 presumably one goes to the settings tab 20:05 sfan5 the other to adv.settings as it says 20:05 sfan5 I fear making advanced settings more prominent will cause users to mess up more 20:06 rubenwardy I don't think duplicate links like that is necessary, people will see the settings icon 20:06 v-rob the "idk something" button should cause their computer to blue screen if they're using windows 20:06 v-rob just for fun 20:07 rubenwardy duplicating links indicates that something is important, and settings aren't that important 20:08 MTDiscord Where are the singleplayer worlds on that layout? 20:08 rubenwardy when you click a game 20:08 MTDiscord OIC, so that's a pre-game landing page 20:08 rubenwardy which are the orange icons in the icon bar 20:08 rubenwardy yeah 20:08 MTDiscord and that bottom left crafting-grid button is the "Go back to the home screen" button? 20:09 rubenwardy yeah 20:09 MTDiscord LGTM 20:10 MTDiscord Presumably there's nothing at all that's strictly necessary to go in the home screen, since all the important shit is in the bottom bar already. So we could probably get something out there given this design pretty quickly, and let the bikeshedding over the exact content happen AFTER we merge the important parts. 20:11 rubenwardy the home screen will at least show Featured Games from ContentDB, allowing us to no longer bundle MTG 20:11 rubenwardy as of yesterday, ContentDB now supports featured packages 20:11 MTDiscord ok so PR when 20:11 hecks here you go you meme goblins https://a.uguu.se/Hfp3p6UN4Fsv_perfection.png 20:11 rubenwardy oh noes 20:11 MTDiscord now it just needs deep fry and warp 20:12 hecks i do not use cancer effects enough to know where they all are. sorry 20:12 Krock needs a picture of a surprised/shouting kid 20:12 hecks no i know 20:14 hecks https://a.uguu.se/xRkCf4j1Vf3y_gonewrong.png 20:15 MTDiscord "We all agreed on a design direction, but unfortunately the work itself was derailed by memes" 20:17 hecks well, if people keep adding bad lighting PRs everything will inevitably turn into free-smiley-de 20:21 MTDiscord https://cdn.discordapp.com/attachments/747163566800633906/868226357220479086/myeyestheyburn.png 20:23 v-rob With graphics designers like hecks, Minetest is doomed. 20:23 MTDiscord always has been 20:23 v-rob good poit 20:24 MTDiscord oh I forgot the fisheye filter 20:24 MTDiscord oh well 20:24 hecks let's stop here before someone takes it seriously 20:25 MTDiscord lol 20:25 hecks with this community, you never know 20:26 MTDiscord oh, that main menu concept was supposed to be a joke? 20:27 hecks it is too late, here comes the PR with 20 upvotes 20:31 v-rob Can we use RTTI in Minetest? 20:31 v-rob I know Irrlicht can't 20:32 rubenwardy https://github.com/minetest/contentdb/issues/321 20:39 sfan5 v-rob: yes 20:39 v-rob cool 20:40 sfan5 though perhaps the cargoculted android flags have -fno-rtti 21:00 sfan5 reminder #11287 21:00 ShadowBot https://github.com/minetest/minetest/issues/11287 -- Take advantage of IrrlichtMt target by JosiahWI 21:01 sfan5 I think we'd do well tagging PRs with "Action / change needed" more consistently when unadressed review comments are left 21:01 v-rob I wish I could review, but I know nothing of CMake 21:02 sfan5 the cmake part isn't so important, more that you can still get mt + irrlichtmt running together in your setup whichever way you currently do it 22:11 sfan5 random reminder that debian has still not packaged Minetest 5.4.0 or 5.4.1 22:11 sfan5 (and with that ubuntu too) 22:12 Pexin they expect you to use snap, right? 22:12 sfan5 ubuntu maybe but not debian 22:24 rubenwardy 20.04 was before 5.4.0, so that's not surprising 22:25 rubenwardy 20.10 or 21.04 should have it though? 22:25 sfan5 "not packaged" = not in repos even if you run unstable 22:26 sfan5 I just read that Debian plans or Debian 11 next month, it will not have 5.4 22:26 sfan5 for* 22:27 sfan5 and neither does 20.10 or 21.04 22:32 sfan5 We should recommend against installing distro versions 22:33 sfan5 in fact I'm gonna PR that to the website repo tomorrow 22:55 MTDiscord Who is the maintainer of the Debian Minetest package? Can we contact them?