Time Nick Message 01:08 schwarzwald[m] https://github.com/guybrush77/rapidobj 01:08 schwarzwald[m] Would it make sense to use this in IrrlichtMt? 06:47 nrz_ whereas it can be interesting, we are C++14 only 10:08 MTDiscord The subset of Wavefront OBJ we actually support is pretty small. It thus doesn't really make sense to drag in importers that implement plenty of features we don't need. I've attached the spec I inferred from reading the IrrlichtMT sources. 10:08 MTDiscord https://cdn.discordapp.com/attachments/747163566800633906/1026435448509300748/irr_obj_spec.md 11:08 appguru (that spec is wrong btw regarding group names: they aren't entirely ignored - only the first name is considered) 11:13 sfan5 @x2048 only broken w/o shaders 11:15 sfan5 GLES is broken in interesting ways https://0x0.st/oJ8w.png 11:23 kilbith the main scene is set in another scene, the wieldmesh 11:24 kilbith quite interesting 12:04 MTDiscord @x2048 12:19 MTDiscord Broken for me with undersampling = 2 and without shaders on nvidia and intel gpu 12:20 MTDiscord Looks like you need to test without shaders too 13:21 pgimeno sfan5: any idea what shader is the culprit? 13:22 pgimeno I mean for the "Irrlicht: GLSL shader failed to compile" error 13:31 MTDiscord What if you set enable_bloom and enable_bloom_debug in your config? 13:32 srifqi sfan5, is it the same as #12824? 13:32 ShadowBot https://github.com/minetest/minetest/issues/12824 -- Shaders won't compile on OpenGL ES 2.0+ after #12791 13:34 pgimeno ah there's a proposed patch already 13:35 vampirefrog there's bloom now? 13:36 vampirefrog okay bloomer 15:06 sfan5 pgimeno: its almost surely the same issue as srifqi linked 16:05 jwmhjwmh Merging #12830 and #12832 in 5m. 16:05 ShadowBot https://github.com/minetest/minetest/issues/12830 -- DevTest: Reject buggy "/hp inf" command by Wuzzy2 16:05 ShadowBot https://github.com/minetest/minetest/issues/12832 -- Disable -ffinite-math-only by TurkeyMcMac 16:06 pgimeno hold on please 16:07 jwmhjwmh OK 16:07 jwmhjwmh Do you need to review them? 16:07 MTDiscord inf or nan should be rejected at a different level 16:07 MTDiscord that is, the engine should refuse modders trying to set hp to inf or nan 16:08 MTDiscord (checking in the chatcommand probably still is a good idea to prevent a crash though) 16:08 pgimeno ok, I was just looking up the reasoning for -ffinite-math-only, go ahead 16:10 pgimeno @luatic yes but Desour has a point, that isinf() returns false with -ffinite-math-only 16:11 MTDiscord I was commenting on the first PR by Wuzzy 16:11 MTDiscord Which should still be merged, but the issues goes deeper and should be solved at a deeper level I suppose 16:12 jwmhjwmh Casting from floating point to integer is par for the course in the Minetest API. 16:12 jwmhjwmh I will merge the PRs then, thanks. 16:13 pgimeno my point was that it can't be solved at the C++ level without a working isinf() 19:41 celeron55 https://forum.minetest.net/viewtopic.php?p=414783 19:42 celeron55 this is interesting from a core dev and/or a distribution perspective 19:43 celeron55 its license appears to respect MT (lgpl2.1) and it's multi platform. so it's worth following 19:45 sfan5 where do the builds come from for platforms that aren't windows? 19:48 celeron55 the posts give the impression that recluse4615 is building non-windows builds on WSL, but i have no idea 19:50 celeron55 anyway, people might have to ready their opinions about having a launcher that's not implemented natively in MT, if that project gains contributors and gets fleshed out 19:56 celeron55 the positive things are probably: it would allow people to contribute who know those technologies rather than C++ and Lua, and allows making a better launcher experience. the launcher is client-side only anyway so it doesn't need to be limited in any way 19:57 celeron55 the negative side is: any work that's put into the launcher stays there only and can't be used elsewhere in MT, and some developers need to have the knowledge to use the different technologies 19:58 celeron55 anyway, i'll be following it to see where it goes and what users think of it 20:01 celeron55 it could be that the launcher portion should be made by a separate team anyway 20:05 celeron55 just need to figure out where to find a team that would work enthusiastically for years to come... on a launcher, of all things 20:17 MTDiscord As long as the launcher isn't mandatory (i.e. MT continues to maintain its own front-end) and there are no lock-in features, I don't see the downsides as being that big of a threat. 20:18 MTDiscord A big upside to 3rd party launchers is that it takes some pressure off of the core team of having to please every possible user. You can allow 3rd party launchers to fill in some niche cases and concentrate resources elsewhere (not that those resources aren't necessarily already elsewhere, but now it's sort of more okay if they are) 20:19 MTDiscord If you're thinking of having an official launcher team, well, all you really have to do is officially endorse some existing launcher once it reaches the standards you have in mind. 20:27 celeron55 that's probably the only way that can possibly work 20:27 celeron55 that's why this is an interesting project 20:27 celeron55 i'm not saying there's a definite or distinct _need_ for it, but it could address some long lasting complaints 20:29 celeron55 and it is an unique opportunity because the interface from a launcher into MT is reasonably well defined 20:29 celeron55 and fairly stable 20:31 MTDiscord With MT's --go command line parameter, it already has been pretty supportive of external "launcher" type solutions, including mine which just parses a file with a list of favorite servers and spits out *.desktop files :-) 20:31 celeron55 as a bonus, a separate launcher can provide features that a launcher built into the MT executable probably will never do, like allowing switching between versions of MT 20:32 celeron55 it seems like fertile enough ground for something generally useful to eventually pop up 20:32 MTDiscord Right, and auto-updating. Alerting the user an update is available is pretty reasonable for MT itself, but trying to actually update the local copy seems pretty hardcore can'o'worms. 20:33 celeron55 well, it's all doable of course. but it may or may not make sense 20:34 MTDiscord Well, some things are just really messy, and it's nice to have somebody else be willing to deal with the mess instead :-D 20:34 celeron55 we'll see 20:38 celeron55 the technology used in this one (tauri + sveltekit) is kind of wild from a minimalist's perspective