Time Nick Message 02:36 MTDiscord I see that the minetest.dustlabs.io has been greatly improved. Fullscreen support, working command and chat keys O.o 02:40 MTDiscord @paradust great job, thanks! Another suggestion: when the game starts, grab the pointer automatically. Detect if esc is pressed, and open the pause menu at the same time (instead of needing multiple esc presses to open the pause menu) also, grab the pointer when the quit command is sent for a formspec. 12:17 rubenwardy web#258 12:17 ShadowBot https://github.com/minetest/minetest.github.io/issues/258 -- Update Education: Add BLOCKALOT, add citation link by rubenwardy 15:11 MTDiscord Hi! I'm having a very weird segfault with an ARM64 (aarch64) build... The arm build crashes with a very weird backtrace too. Is there any specific flags i should use to build arm64? I'm building inside a Docker container using the same flags and a Github action for multiarch cross compiling. The runtime is also Docker on a VM. The x86_64 build works ok, no random segfaults, but on ARM it does. Both are from the 5.5.1 source. Here is a 15:11 MTDiscord sample segfault extracted from Debian 11 gdb on the container: https://pastebin.com/srjRAmck 15:11 sfan5 using luajit? 15:11 MTDiscord It seems to crash on start, for about three times in a row, then works for a while and crashes again. 15:11 MTDiscord sfan5: let me check. 15:11 MTDiscord (it sometimes crashes during startup too) 15:12 MTDiscord I'm almost sure it is something wrong in my host config but I'm not finding out what it is 15:12 MTDiscord Yes, should be using luajit on both arch builds: https://github.com/ronoaldo/minetestserver/blob/main/Dockerfile#L13 15:17 sfan5 if it's aarch64 then you'll want to build LuaJIT from git source, the latest "release" is from 2017 and what debian ships. there are known issues with it 15:18 MTDiscord got it, thanks 15:28 MTDiscord sfan5: just so I can learn a bit, if you don't mind sharing - how did you figured from the backtrace it is be related to the lua part? 15:30 sfan5 JIT engines dynamically generate code, unlike other code it won't have any symbols at all nor a corresponding binary, so generally if you see '?? ()' in the backtrace that's bound to be luajit 15:31 sfan5 in this case however the stack is corrupted, since luajit deals with more low-level things and such (and also from experience, knowing that such issues have happened before) it's more likely that luajit is involved rather than e.g. a compiler bug 15:32 MTDiscord nice, thanks for sharing the hints! i'll try rebuild with a more recent luajit ? 15:50 sfan5 appguru: can you check whether minetest.punch_node causes the global on_punchnode callback to be called? 15:51 appguru sfan5: in the unit test? should I hook it? 15:51 sfan5 I'm wondering in general but if you write the code anyway you can also put it in the unit test ;) 15:52 appguru ugh, this is dirty: "bool success = scriptIfaceNode->node_on_punch(pos, n, NULL, PointedThing());" 15:53 appguru that's why it crashed: it's passed a type="nothing" pointed thing 15:53 sfan5 oh also if you're there anyway edit the comment, the part about a "not funtioning objectref" is no longer correct 15:53 appguru the docs don't mention that puncher can be nil in on_punch 15:53 sfan5 it gets passed `nil` 15:54 sfan5 or actually does it 15:54 sfan5 hm 15:58 sfan5 if you have more data to verify you can look into entity.lua for inspiration but here a bool variable should suffice 15:59 sfan5 ah nope the objectref thing is still accurate, the thing I recently changed was on_death and only that 16:09 sfan5 appguru: you're missing something but also I meant minetest.register_on_punchnode 16:32 rubenwardy merging web#258 in 10 16:32 ShadowBot https://github.com/minetest/minetest.github.io/issues/258 -- Update Education: Add BLOCKALOT, add citation link by rubenwardy 16:34 kilbith I'm surprised there's no mention of Kidscode 17:00 sfan5 appguru: still broken :) 17:00 appguru yeah I see 17:00 appguru and I wonder why 17:02 appguru when I start devtest with unittest auto-running enabled everything is fine 17:02 sfan5 ./util/test_multiplayer.sh is the headless environment 17:02 appguru yes 17:02 sfan5 also speaking of deprecated vector.new 17:02 sfan5 local pos_copy = vector.new(pos) 17:02 appguru but why wouldn't on_punchnode work in the headless environment? 17:07 appguru sfan5: i kinda wanted to test the entire something_player_action_node family, should've renamed the unit test appropriately 17:08 sfan5 on_punch calling punchnode as a fallback seems kinda badly designed 17:08 sfan5 I'd expect register_on_punchnode to run first and then nodedef.on_punch 17:09 sfan5 I suppose nobody redefines on_punch for nodes so that's not noticed 17:10 appguru should I stuff more deprecated vector.new removals into the regression fix PR? 17:10 sfan5 fwiw on_dig works the same and people know how to use that correctly (or do they?) 17:10 sfan5 if you rename it, sure 17:10 appguru I don't think any mods handle pt.type = "nothing" correctly :) 17:12 sfan5 one reason could be that the map is not loaded but that'd be a major fail of the test framework 17:12 sfan5 have fun debugging this 17:27 appguru sfan5, on an unrelated note, do your win builds have ncurses? 17:32 sfan5 no 17:32 sfan5 I believe it doesn't even work on windows 17:36 MTDiscord Would that be due to how Minetest uses it? I've had no issues with my C+ncurses projects for Windows/Linux 17:41 Krock I believe the best compatibility would be reached through the cygwin/msys2 terminal 17:42 Krock not sure if the normal cmd even supports colors 17:46 appguru sfan5: i'll just revert the unit test I guess 17:46 appguru or revert the check for whether on_placenode is called 17:46 sfan5 or comment it out 17:47 appguru outcommented it is ^.^ 17:48 appguru going on a vacation starting tomorrow 17:54 sfan5 hopefully in a place with cooler weather 18:06 sfan5 by the way it would be nice if someone could review #12418 #12505 since both make sense for 5.6 18:06 ShadowBot https://github.com/minetest/minetest/issues/12418 -- Remove float vector range checks from scriptapi by sfan5 18:06 ShadowBot https://github.com/minetest/minetest/issues/12505 -- More corrections to lua_api.txt by sfan5 18:19 sfan5 kilbith: I looked at #10100 and it has some cool stuff but it doesn't seem like the usecases are that broad 18:19 ShadowBot https://github.com/minetest/minetest/issues/10100 -- Additional texture modifiers by Treer 18:20 sfan5 direly neeeded for example would be a [combine that operates (and resizes?) with percentual texture sizes, perhaps with some enforced minimum 18:20 sfan5 so that you can operate on textures with unknown size 18:35 Krock will merge #12519 #12527 #12539 in 15 minutes 18:35 ShadowBot https://github.com/minetest/minetest/issues/12519 -- Deserialization: Restore backwards compat by appgurueu 18:35 ShadowBot https://github.com/minetest/minetest/issues/12527 -- GUIFormSpecMenu: Fix multiline translation by SmallJoker 18:35 ShadowBot https://github.com/minetest/minetest/issues/12539 -- Fix regression & replace more occurrences of vector.new with vector.copy by appgurueu 18:36 sfan5 will merge #12418 after that 18:36 ShadowBot https://github.com/minetest/minetest/issues/12418 -- Remove float vector range checks from scriptapi by sfan5 18:49 Krock merging.... https://media.giphy.com/media/cFkiFMDg3iFoI/giphy.gif 18:51 Krock done. ping sfan5 18:58 rubenwardy haha, so the recursive dependency algorithm also tries to enable game mods 19:02 sfan5 how did you introduce this regression? could there be more? 19:03 rubenwardy I didn't 19:03 sfan5 oh 19:03 sfan5 how come then 19:03 rubenwardy it was already the case, it `.enabled` just wasn't being read 19:03 rubenwardy but with my new code, I was passing all mods with .enabled = true to the mod config class 19:04 sfan5 shouldn't you get rid of setting .enable instead of mending this in the display code? 19:04 rubenwardy I did 19:04 rubenwardy https://github.com/minetest/minetest/pull/12284/commits/24b175349ca7b44d6348ae26f9d6ac6d8abcd347#diff-a400b0be5dbda5a4d8f53e6c6a7877ed36fe670e805f0b5932724eb7c321a1ccL545 19:04 sfan5 whats dlg_config_world.lua for then? 19:05 rubenwardy that change isn't needed 19:06 rubenwardy I'll undo that bit 19:12 sfan5 btw does anyone have an opinion on irr#120 (functionally)? 19:12 ShadowBot https://github.com/minetest/irrlicht/issues/120 -- CGUITabControl: Center selected tab whenever possible by SmallJoker 19:56 Krock it's really an edge-case use for formspecs with many tabs. low priority. 20:55 rubenwardy merging #12284 in 10 20:55 ShadowBot https://github.com/minetest/minetest/issues/12284 -- [No Squash] Show dep errors in Select Mods modal by rubenwardy 21:05 MTDiscord is 5.6.0 feature freeze happening this weekend? 21:09 sfan5 I don't see why not 21:11 Krock yes, planned on sunday. 21:20 ROllerozxa so I assume the android crosshair PR is gonna have to wait yet another version considering it's not in the 5.6.0 milestone? 21:20 sfan5 how simple is it? 21:22 ROllerozxa well it's already gotten one approval. #7865 21:22 ShadowBot https://github.com/minetest/minetest/issues/7865 -- Add crosshair support for Android by srifqi 21:26 rubenwardy it was a lot simpler than I expected 21:27 MTDiscord 'tis a very nice addition, probably worth getting in asap 21:36 ROllerozxa I'd actually want to see it even be enabled by default but I don't know what the opinion of changing default controls like that are, having it as an option is better than nothing I guess 22:36 erle sfan5 i would indeed appreciate a [combine with percentual texture sizes. it is necessary for texture wraparound with nodeboxes that exceed the bounds of a node. 22:37 erle like, when you make a nodebox that looks like a cube that is shifted by 1/16 down so it can visually sit on a 15/16 nodebox 22:37 erle then you need to adjust the textures 22:38 erle and one way to do it is combine them with shifted/cropped versions of themselves 22:38 erle in the end i settled for making it work for 16×16 textures only 23:49 erle sso we talked about shaders at some point. i figured out you can just dump them by means of mesa environment variables: https://docs.mesa3d.org/shading.html#envvars