00:00:16 i think it just wants you to remove a tab character at the minus 00:00:32 That's probably it. 00:00:38 Question is, why? 00:00:44 because its an empty line 00:00:52 Is that little tab hurting anyone? 00:01:00 it hurts git 00:01:06 it makes patches hard to read 00:01:12 Ah, fair enough. 00:01:20 Alright, I'll go kill some tabs. 00:01:38 ```\t\n``` is on each line 00:01:46 invisibly 00:01:47 lol 00:02:50 when you are done you need to git rebase --interactive on the master branch and squash into a single commit, but when you get to it working let me know and I can talk you through it 00:02:59 although its midnight here, so I will be asleep 00:03:06 until tomorrow 00:03:11 im sure someone can explain rebasing properly 00:03:17 if not google a youtube tutorial 00:03:21 we do it all the time 00:03:26 so its very useful knowledge 00:03:28 but beware 00:03:35 backup before you do it 00:03:54 it rewrites your history, if it goes wrong and you force push 00:04:02 git has a 'distributed' broken copy XD 00:04:09 so no backup 00:04:38 git kraken makes the interactive rebase step simpler 00:04:48 then once you do it once in there 00:04:51 try out the command line :) 00:05:14 OK im out 00:05:16 :zzz 00:05:21 💤 00:05:24 goodnight 00:46:04 reduz clayjohn if i have a plugin that i want to render an infinite grid in the editor viewport what's the best way to do that? 00:47:11 jonbonazza[m]: is it for a specific node? 00:47:54 Yes when one of my custom nodes is selected in the scenetree 00:48:22 Basically like what GridMap is doing 01:30:07 I finally got it working. Yay! No more crazy complaints from GitHub! 01:57:17 Syntaxxor: Your code is in a good state (roughly speaking, I haven't reviewed it in detail), but you still need to squash it down to one commit (or another small number, but don't worry about that, we will aim for one). Where it says "Commits" at the top, it should say "1" instead of "13". 02:10:32 Ah, good to know. 02:30:15 Alright, managed to do that. 03:24:42 Akien: I wrote timestamps in a comment in the Q&A video, feel free to copy them into the description https://www.youtube.com/watch?v=xegEjjKjJgQ 04:51:24 fales if you need help with the stun server and the networkpeer changes needed to support it let me know 06:06:27 Calinou: if you are around, I have a proposal I want to write, but I have a suspicion that it might already exist. However, I am unable to think of what to search to check and I can't really find it in the backlog. 06:07:08 My proposal is to provide a way to mark custom class names as "not exposed" in the editor, so that they don't show up under the Add Node dialog 06:07:51 This is most useful for libraries or frameworks where you want the user to use a scene instead of a script directly, but that scene has a script attached with a class_name of the same name as the scene. 06:08:09 Are you immediately aware of such a proposal already? 06:08:41 Could be a good use of annotations in 4.0 06:08:50 @vne 06:08:56 vnen: for your consideration as well 06:37:55 Is there a reason that `node["property_name_here"]` works for setting in 4.0 but not getting? It works for both in 3.2.3. 06:38:27 In the current master branch you have to do `node.get("property_name_here")` 06:40:57 If the `[]` is being purposefully removed as a shortcut for getting, why is setting still possible? 06:42:08 Morning 06:42:49 aaronfranke: awesome, thanks for the timestamps. I'll copy them as soon as I've brewed some coffee :D 06:43:35 As for [] to get/set member variables, I didn't even know that this was possible :P but yes if only one works it's a bug. 06:44:37 jonbonazza: doesn't ring a bell but I'm also not up to date on all proposals. Makes sense though, could be added via an annotation I guess. 07:01:28 Wii 07:14:27 vnen: I'm not sure if a bug or a limitation of the current gdscript implementation, but if I have a class called `BattleCreature` I can't reference the class name in that script. 07:14:51 So for instance, I can't check if another node is an instance of BattleCreature within BattleCreature's script. 07:15:00 It gives a cyclic refernence error. 07:15:04 reference* 09:15:51 jonbonazza[m]: I don't remember seeing a proposal about that, feel free to create one. Sounds like a good use case for annotations indeed 09:16:51 also, the `master` branch already has an infinite 3D grid which might be backportable to 3.2.x 09:16:55 https://github.com/godotengine/godot/pull/28289 09:32:37 Calinou: FYI: https://github.com/SCons/scons/pull/3810 09:32:50 So we can finally drop any reference to pywin32 once 4.0.2 is released I guess. 09:33:05 yaay 09:35:23 In general we should probably tweak our compiling pages to always recommend using the very latest version of each dependency, and only mention the min supported one afterwards. 09:36:42 Currently we have e.g. this: "SCons 3.0 build system. If using Visual Studio 2019, you need at least SCons 3.1.1." 09:36:54 definitely, my Linux dependencies installation script installs SCons from pip for this reason 09:37:06 (so you don't get an outdated version from the distro) 09:37:29 This should likely be "Latest SCons version (e.g. from `pip`). The minimum supported version is SCons 3.1.1." 09:38:24 SCons performance and reliability on incremental builds seems to have increased significantly in 4.0 so we should really encourage users to have the latest one. 09:38:49 And yeah we should probably recommend `pip` installation of scons for everyone. Might be worth putting in the Introduction to the buildsystem doc, and then link to it from all OS-specific pages. 10:18:34 yay, I backported https://github.com/godotengine/godot/pull/28289 to 3.2 and it worked on the first try 10:19:28 grid transparency now works as well 10:23:45 https://github.com/godotengine/godot/pull/43206 10:30:26 Nice :) 10:30:46 vnen: Yeah I can reproduce the endless linking issue with VS 2019 for tools=no target=release. 10:31:39 vnen: `gdscript_vm.windows.opt.64.obj` is 0 KB so I guess it's taking forever to generate this file, which is the last step before linking the .lib for modules/gdscript and then the main Godot binary. 10:54:04 morning 10:56:48 jonbonazza[m]: it's a limitation in 3.x. In master it should be fixed though I received reports that it wasn't working (but will be fixed before 4.0) 10:59:03 I also backported a shadow rendering fix from https://github.com/godotengine/godot/pull/37678 10:59:14 it's definitely an improvement over 3.2, need to see how it compares with master 10:59:28 moorning 10:59:37 Akien: yeah, took me 3h to build this file. Not sure why it take so long to compile this. I'll see if there's a way to measure the bottleneck (I've found some ways to debug build time but have not tried them yet) 13:31:31 RandomShaper: so I guess we can merge #43049, and open an issue about the bug in master/3.2? 13:31:31 #43049: Fix script used as type gone too early (3.2) | https://git.io/JTVQX 13:32:00 (and make that bug high priority as I'd hope to have it fixed before 3.2.4 beta 2) 14:20:55 Akien, yes. And if vnen or someone else doesn't come up with a solution for the remaining bug, I'll do it. 14:21:18 RevoluPowered: https://www.phoronix.com/scan.php?page=news_item&px=RadeonSI-Uber-Inline-Uniforms 14:21:18 > The goal with this is to improve the performance for uber shaders, or rather large shaders with lots of possible branches at run-time. 14:21:24 sounds like this will benefit Godot 4.0 14:23:40 did you mean reduz? : D 14:28:58 Calinou: possibly, but I have no idea what this means 14:29:09 Calinou: I will eventually have to work with the IHVs on optimization 14:30:59 RandomShaper: thanks, I merged the PR. Would you have time to open an issue to track the issue you and @ChainedLupine have been debugging? 14:31:13 I can do it too but I guess you'd know how to phrase it more accurately. 14:39:21 Akien, sure. I'll do it in the new hours. 14:39:52 Thanks 14:41:39 halfway there.. https://usercontent.irccloud-cdn.com/file/7DbVe777/image.png 15:03:36 Nice :D I guess Calinou or kobewi can send you a blue version of the DirectionalLight3D and PointLight3D to use :D 15:03:51 It's funny how companies always schedule their tweets to go out at the exact same time: https://snipboard.io/7rjpuN.jpg 15:04:05 Proof that Godot is not corporate, I never schedule @godotengine posts ahead :P 15:14:48 is there an existing proposal for custom annotations? 15:24:49 nm, i found it: https://github.com/godotengine/godot-proposals/issues/1316 15:47:08 RevoluPowered: no, I meant you since you have an AMD card with the open source driver 15:48:07 reduz: I wonder if we should name it OmniLight2D instead, since technically, a 2D point light is always omnidirectional, even if you use a texture that looks like a spotlight 15:48:20 with OmniLight3D, you can simulate spotlights with projector textures now so it's analogous to that 15:48:51 (it's useless, but you can do it :P) 15:50:47 Oh yeah I am hoping the new release of the AMD stack and linux will help the AMD related problem 15:51:14 TMM: told me that there was an LLVM bug causing havok with the sound driver built into the card and causing some of the power play problems 15:52:16 the snd_intel_sth was used in the amd cards for the HDMI audio and in some cases the negotiation on init() 15:52:28 so causing the power play problems 15:52:42 I hope that linux will be pushed soon with the new llvm fixes 15:52:50 Yeah, it just caused the gpu to freak out, and thus the hdmi/dp audio section as well 16:04:25 ifire: we now know what is broken with the sensei rig btw 16:04:55 someone at work discovered its when you check 'remove unused vertex influences' because they had the same bug with a new mesh :) 16:05:20 the ERR_FAIL_COND with sbs > sb I mean 16:06:28 Calinou: maybe, OmniLight has a radius though, but this one is a single point and a texture 16:06:52 Calinou: if they were named the same, I think users might expect them to work the same,but they dont 16:22:16 how do i re-trigger CI? 16:22:27 i rebased a branch but CI didn't run :( 16:23:14 you can close and reopen the PR, or amend a commit with no changes and force-push it 16:23:23 ok ty 16:35:04 how do I create canvas editor plugin, like e.g. TileMap? is it possible? 16:42:21 ok found it 16:53:46 so I discovered that make_visible(false) is not called when the edited object is deselected. Is this intended? I once opened an issue that spawned because of this 16:58:13 yeah, gizmos tend to linger around after you deselect a node in both the 2D and 3D editors 16:58:19 I have no idea how to fix it, though 17:03:29 congratulations on the fbx merge RevoluPowered! my twitter is blowing uuuuuup 17:03:55 Mixamo should thank you. :P 17:13:44 haha xD they should 17:13:48 Autodesk should thank me 17:13:50 and take the importer 17:13:52 and dump the sdk 17:13:54 and hire me 17:14:10 SDK is so fucking slow 17:14:14 someone who works at Autodesk was on the Discord server a few weeks ago 17:14:32 spy 17:14:34 I was just kidding 17:14:44 jfons: ive tried out your light mapper it seems stable enough for the beta, i was just wondering if there was anything left you have to do before that. Last week you mentioned upgrading embree, it seems thats been done. 17:14:50 I like working for IMVU and Prehensile 17:16:04 congrats on fbx, that will be useful since like half the model that i buy are in fbx 17:16:41 Need to consider adding `modules_enabled_by_default` in SCons: #41091 17:16:42 #41091: Add option "modules_enabled_by_default" for shorthand disabling modules | https://git.io/JJPON 17:17:21 Because now with fbx merged, I have to remove assimp manually and disable fbx manually with `module_fbx_enabled=no`. :D 17:17:25 But congrats! 17:24:57 OK first draft I will expand more as I think of more 17:24:58 https://docs.google.com/document/d/1wMl67XcgaoliccdTznu1SPk45tnYem2Q-Htg-j6nvEg/edit?usp=sharing 17:30:03 would the fbx stuff be able to be used in assimp? 17:30:10 *upstreamed to them 17:31:09 one day 17:31:14 maybe 17:31:23 the problem is they need to throw away a lot of their work 17:31:33 their aiPostProcesses can't work with FBX 17:31:48 its too many layers 17:31:51 its a bad idea 17:31:57 we learned that the hard way with a rewrite 17:32:08 Looks good RevoluPowered! The only thing I might change, if I was going to change anything, is breaking point 1 in "Why did you not use the FBX SDK" into two points: one for the size of the SDK and another for it not having source available. Outside of that, it looks good to me! 17:32:09 Because with Assimp you have an additional format to convert 17:37:15 ```We can add proper support for complex features like Segment Scale Compensation, which unity currently does not support.``` 17:37:22 is this a burn 17:37:23 or 17:37:27 how do i say it nicer 17:37:27 xD 17:37:29 🔥 17:38:12 twisted_twigleg[: simplified! 17:38:14 thanks 17:38:16 good spot 17:38:33 No problem! Honestly, it was probably fine before, I just thought it might be better as two points. 17:39:25 With the complex feature support, is Unity the only engine that does not support it? Or is it a feature generally not supported? If the latter, I might change it to "... which other game engines currently do not support" or something like that so it's not calling Unity (or any game engine) in specific out for their lack of support. 17:46:25 It's unity doesn't support it 17:46:30 and maya has a post about how to disable it 17:46:34 explicitly mentioning unity 17:46:35 :P 17:47:02 Well, if its good enough for Maya, its probably fine to mention Unity specifically XD 17:47:18 Revision 2 is up 17:47:22 with grammar fixes from Grammarly 17:47:27 it reports it is correct :D 17:48:03 I'll make it a bit more gripping at some point tomorrow maybe 17:48:08 unless we need to do a blog post today 17:50:26 Looks good to me! 17:58:52 RevoluPowered: no hurry we can publish tomorrow 17:59:02 RevoluPowered: whoo 18:01:05 is @NodrogMrDev Gordon? 18:03:14 follows 18:04:36 i need a fast segment-rectangle intersection 18:04:58 NHodgesVFX: I did upgrade embree, and then I ran a profiler over the whole lightmapper and optimized a bunch of little things :) 18:05:22 I am doing some extra tests now, but I hope to be done this weekend (fingers crossed) 18:06:03 jfons: do you know where in the source code is the Triangle BVH4 vs line segment intersection code? I could not find it 18:06:20 (of embree) 18:08:40 jfons: very cool look forward to it being in 3.2.4, seems like you might be done for beta 2, if not there's always beta3+ 18:08:52 reduz: not sure myself either, embree has a bunch of abstraction layers that make it hard to navigate 18:09:15 NHodgesVFX: yup, I really want to get it done before beta2 18:25:25 jfons: is this one gonna need the changes we had to do to make the mesh format to come through properly? 18:25:52 we had to make SurfaceTool in our engine initialise the format properly and there was some other fixes we did 18:26:45 I don't use twitter so much anymore iFire, but maybe will in the future it's just nice to see users wanting the functionality so much 18:31:54 reduz: can you tag odino: and ernest in the tweet? 18:31:59 iFire 18:32:15 RevoluPowered: nope, it's something unrelated :) 18:32:19 https://twitter.com/iFiery 18:32:52 RevoluPowered: ah i cant edit the twet, add much add an answer, but make sure to properly credit them in the article and the godotengine account will tweet this properly tomorrow 18:33:13 ok cool to reduz and joan :) 18:33:17 same answer for two things 18:33:18 :D 18:33:49 Twitter said we can have an edit button when everyone wears a mask 18:43:46 ha 18:46:37 iFire: can you tell that I used a bot back in 2016 18:46:48 on twitter 18:46:49 xD 18:47:01 I kept being stuck at 100 followers 18:47:28 i removed it after a year 18:47:42 i had to unfollow so many weird accounts.. would not recommend. 18:47:57 RevoluPowered: my friend told me to post daily 18:48:09 and the followers will arrive 18:48:26 yeah definitely 18:51:55 my Twitter only serves as a news source for what I do, nothing else really 18:52:02 I still have ~1.4k followers 18:56:48 nice 18:56:56 I don't do much on any of them anymor 18:56:58 anymore 19:16:09 reduz: Calinou I started doing a comparison of ASSAO and CACAO last night. ASSAO is indeed much faster on intel devices. I wonder how they compare on NVidia devices. 19:16:57 I think I can optimize ASSAO using compute as well. Right now it does a lot of small draw calls. So should be able to come up with something nice. 19:18:23 I realized why I liked CACAO more initially, it handles a large radius way better. With ASSAO, if you make the radius any larger, the quality goes way down. With CACAO, it always looks a little blurrier, but it doesn't look any worse when you increase the radius. 19:24:30 clayjohn: yeah thats right, although the falloff in ASSAO is much nicer and feels a lot more natural than in CACAO, which just ends abruptly 19:25:04 clayjohn: which one is simpler code wise? 20:04:06 Hello guys! I'm trying to help with solving an issue with get_joy_axis that it's skipping steps. A friend tried to replicate the issue on MacOS (I'm on Windows) and he's experiencing the same issue but in larger steps. 20:04:20 Here's the issue: https://github.com/godotengine/godot/issues/42876 20:07:59 Where do you think I should start looking into? What's the root class? 20:12:56 nvm I found it, going to start checking at Input.cpp 20:24:18 Anyone knows where I can find the method "is_valid"? 20:24:48 https://github.com/godotengine/godot/blob/9dad483920ffa3c32a03f1f4cc869b3f7d1cbe0e/core/input/input.cpp#L586 20:25:42 is_valid() is equivalent to !is_null() 20:25:51 it's part of Object IIRC 20:26:04 oh thx 20:26:37 to find where a method is declared, you can usually search for "::is_valid(" in the whole codebase 20:26:48 it won't always work but is good enough 99% of the time :P 20:27:54 Yeah I'm just browsing it in github, how could I do that search btw? 20:28:29 I'll clone it so I can do that 20:31:26 GitHub's code search isn't reliable enough to do that, unfortunately 20:31:31 cloning the repository is often a better approach 20:33:06 Do you know the shortcut to search in the whole codebase, in VisualStudio? 20:33:31 Or was it ctrl+click to see the definition? 20:33:42 It's been a long time hah 20:36:29 I want to fix this just in case someone wants to make a racing game or something like that, since it will halve the wheel precision at least 20:36:56 and with those you can't flash a firmware to the PCB haha 20:38:01 I use VS Code, not Visual Studio 20:38:04 it's Ctrl + Shift + F here 20:38:35 alternatively, if you know your way around the command line, install https://github.com/BurntSushi/ripgrep and use that :) 20:46:38 reduz: I think ASSAO will be simpler. But I haven't looked in detail. CACAO provides glsl and Vulkan code which is nice. ASSAO is fully directx so it will be more be more work to translate. 21:12:11 clayjohn: sounds great 23:56:53 Calinou, since you mentioned automatic PascalCase conversion a couple of times now, I wanted to hint you to a PR where I did that for the editor and got some feedback. Maybe it is worth a consideration in your plans: #40631 23:56:54 #40631: Add C# Editor Documentation | https://git.io/JJ83e 23:57:33 UI-wise, I think your implementation is fine 23:57:48 toggling between syntax isn't something you'll do very often, so an editor setting seems alright to me 23:57:59 (maybe a project setting to set a "preferred" language could be used in the future, still) 23:58:05 yeah, I am more speaking of https://github.com/godotengine/godot/pull/40631#issuecomment-663876898 23:58:32 I already introduced one language setting for the examples 23:59:14 Which defaults to GDScript in normal builds and GDscript and C# in mono builds 23:59:17 I can't speak much about the API generation part itself, I have little experience with Mono in general 23:59:37 is that language setting a project setting or an editor setting? 23:59:49 if it's a project setting, be careful as it'll vary across machines if no value is set in the project.godot file 23:59:51 editor setting 23:59:56 ah, that should be fine then