Minetest logo

IRC log for #minetest-dev, 2022-07-10

| Channels | #minetest-dev index | Today | | Google Search | Plaintext

All times shown according to UTC.

Time Nick Message
00:18 behalebabo joined #minetest-dev
00:28 kilbith joined #minetest-dev
01:09 independent56 joined #minetest-dev
01:28 independent56 joined #minetest-dev
01:32 independent56 joined #minetest-dev
01:34 independent56 joined #minetest-dev
01:36 independent56 joined #minetest-dev
03:18 olliy joined #minetest-dev
04:00 MTDiscord joined #minetest-dev
04:28 Yad joined #minetest-dev
05:55 calcul0n_ joined #minetest-dev
09:04 Warr1024 joined #minetest-dev
09:10 YuGiOhJCJ joined #minetest-dev
09:25 Fixer joined #minetest-dev
09:26 independent56 joined #minetest-dev
10:11 YuGiOhJCJ joined #minetest-dev
10:15 HuguesRoss joined #minetest-dev
10:51 Yad joined #minetest-dev
10:56 appguru joined #minetest-dev
11:10 independent56 joined #minetest-dev
11:19 Yad joined #minetest-dev
11:57 Taoki joined #minetest-dev
11:58 Yad joined #minetest-dev
12:27 Warr1024 joined #minetest-dev
12:41 kilbith this commit causes a regression with the sound in formspecs: https://github.com/minetest/minetest/commit/e51f47461
12:42 kilbith I cannot hear the click sound on i3 anymore (buttons use the `sound` property in StyleSpec)
12:42 kilbith ^ Krock
12:46 MTDiscord <luatic> kilbith: Open an issue.
13:23 Krock hmm
13:25 Krock I see. it's an awful implicit type conversion
13:26 Krock C++ likes to cast "false" to 1.0f
13:26 Krock *  0.0f  but you get th idea
13:35 Krock kilbith: please confirm whether this completely restores the previous behaviour: https://krock-works.uk.to/u/patches/0001-GUIFormSpecMenu-Fix-parameter-order.patch
13:35 Krock I tested it locally and it works well from what I can tell
13:38 kilbith Krock: seems to work
13:38 Krock thanks. will push in 15 minutes
13:53 Krock pushing
14:11 rubenwardy Krock: Minetest's GUISkin class is needed to support formspec styling
14:11 rubenwardy it could be upstreamed into irr-mt though
14:16 kilbith > CDB has 1488 packages, with a total of 6000872 downloads.
14:16 kilbith 6M reached
14:17 MTDiscord <luatic> ?
14:21 sfan5 rubenwardy: other way around rather
14:31 Krock remove CGUISkin from Irrlicht
14:49 HuguesRoss joined #minetest-dev
15:00 HuguesRoss joined #minetest-dev
15:00 Yad joined #minetest-dev
15:07 HuguesRoss8 joined #minetest-dev
15:12 HuguesRoss3 joined #minetest-dev
15:21 kilbith http://codepad.org/RgfDjFKE
15:22 kilbith I'm not receiving anything from GenericCAO, I am missing something?
15:22 kilbith network really isn't my thing
15:31 Yad joined #minetest-dev
15:43 kilbith yeah I miss sending a packet from UnitSAO
15:45 Yad joined #minetest-dev
15:47 Yad joined #minetest-dev
15:50 Yad joined #minetest-dev
15:55 fluxionary joined #minetest-dev
15:56 kilbith UnitSAO::attachCamera (ID): 1
15:56 kilbith got it
16:11 olliy1or joined #minetest-dev
16:16 fluxionary a couple days ago, someone suggested building minetest with "sanitizers", so i built a client w/ memory sanitizers, and it found some (not huge) memory leaks in irrlicht. is anyone actually interested in that output?
16:21 fluxionary er, "address" sanitizers
17:32 olliy joined #minetest-dev
18:03 erle fluxionary was it me? ;)
18:04 erle fluxionary i suggest to open issues on the bug tracker, but look before if there are already ones. if it applies to upstream irrlicht, open issues there as well.
18:10 fluxionary alright, created https://github.com/minetest/irrlicht/issues/121.
18:11 Warr1024 joined #minetest-dev
18:11 fluxionary running w/ memory sanitizers might be more productive (and compiling luajit w/ those too), though i'll have to figure out how to get things to recognize clang (gcc doesn't support that)
18:26 Krock at least valgrind gets a bit confused about LuaJIT sometimes
18:27 Krock I did the memory sanitizer thing for Minetest a while ago. there's indeed a few minor issues, but generally not too concerning
18:49 schwarzwald[m] fluxionary: `CMAKE_CXX_COMPILER=clang CMAKE_C_COMPILER=clang` environment variables do the trick IIRC. Been a little while since I did it though.
18:52 schwarzwald[m] Requires clean build directory to set compilers (can't change it for existing build) so out of source build is probably a good idea.
18:52 fluxionary ah nice, i noticed `CC=clang;CXX=clang++` didn't work and went back to doing other things
18:53 sfan5 $CC and $CXX do work, trying to set CMAKE_ stuff via environment might not
18:53 independent56 joined #minetest-dev
18:53 fluxionary gotta reboot, back in a moment (hopefully)
18:55 fluxionary joined #minetest-dev
18:57 sfan5 most of that log (where it says "Indirect leak" I guess) is probably stuff that just happens to be around at shutdown and wasn't deleted but could still be
18:57 fluxionary aha. what about the direct leak stuff?
18:58 sfan5 that probably a problem
18:59 sfan5 +'s
19:00 fluxionary fortunately none of the leaks seemed very big, though it'd be interesting to see what happens on a busy server
19:02 fluxionary (which obviously wouldn't have irrlicht stuff going on)
19:08 fluxionary `CMAKE_CXX_COMPILER=clang` (and C) worked when passed as an a `-D` argument to cmake, though the build failed w/ a linker error... hm.
19:13 sfan5 it's clang++
19:21 independent56 joined #minetest-dev
19:23 fluxionary either one works i think
19:24 fluxionary in any event, it looks like using memory sanitizers requires building custom versions of the standard libraries...
19:24 sfan5 clang will "works" but miss libstdc++ in the linking step
19:25 sfan5 hence clang++, which is meant to compile c++
19:27 fluxionary oh, that was my linking issue
19:27 fluxionary in any event, i've got more work to do if i want to try this feature
20:12 Baytuch joined #minetest-dev
20:22 fluxionary gave up on that for now, maybe i'll try it another day
20:57 erle joined #minetest-dev
21:57 kilbith Some exception: "vector::_M_range_check: __n (which is 0) >= this->size() (which is 0)"
21:57 kilbith does it make any sense?
21:58 kilbith it throws an exception on .at(i)
22:08 schwarzwald[m] kilbith: Indexing the vector out of bounds, but maybe you were asking in a more specific context.
22:09 kilbith I can understand the > part which is indeed out of bound but what about the = ?
22:12 schwarzwald[m] >= because at(size()) would be out of bounds.
22:13 schwarzwald[m] Size 0 means so elements so index 0 is >= size and therefore out of bounds.
22:13 schwarzwald[m] s/so/no/
22:14 schwarzwald[m] Sorry, I forgot editing a message sends a new copy to IRCE
22:14 kilbith oh I thought that vector.size() == 0 when it was filled with one element
22:15 kilbith got confused vector.at(0) accessing the first element
22:34 panwolfram joined #minetest-dev
22:42 erle kilbith oh, are there any languages where that is indeed the case that you always have to remove 1 to know the size?
23:31 rubenwardy MacOS bugs: https://appdot.net/@mdhughes/108624993095436787
23:33 rubenwardy do any core devs have a mac?
23:40 MTDiscord <Jonathon> dead link btw
23:40 rubenwardy works for me
23:41 MTDiscord <Jonathon> maybe you need to be signed in?
23:41 rubenwardy I'm not signed in
23:41 MTDiscord <Jonathon> odd
23:42 rubenwardy https://rwdy.uk/hM1R1.png
23:56 MTDiscord <Warr1024> I've had those Mac input lag problems before.  What I observed wasn't a fixed lag kind of thing, it felt more like inputs were being buffered faster than they could be processed, so if I held the dig button down longer, it would keep swinging longer after I tried to stop.
23:58 MTDiscord <Warr1024> I actually got used to ending a dig by pointing off somewhere that wouldn't cause damage while I waited for the tool to calm itself back down.  I just assumed it was part of the "nobody really cares about Mac" problem.

| Channels | #minetest-dev index | Today | | Google Search | Plaintext