Minetest logo

IRC log for #minetest-dev, 2018-06-10

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

All times shown according to UTC.

Time Nick Message
00:01 Fixer joined #minetest-dev
00:27 Taoki joined #minetest-dev
00:32 lumberJ joined #minetest-dev
00:42 YuGiOhJCJ joined #minetest-dev
01:11 p_gimeno joined #minetest-dev
01:27 ANAND joined #minetest-dev
02:15 Tmanyo joined #minetest-dev
02:27 entuland joined #minetest-dev
02:37 AndroBuilder joined #minetest-dev
04:11 paramat joined #minetest-dev
04:54 VargaD joined #minetest-dev
05:39 VargaD joined #minetest-dev
06:07 VargaD joined #minetest-dev
07:41 YuGiOhJCJ joined #minetest-dev
08:00 Gael-de-Sailly joined #minetest-dev
08:33 sfan5 merging #7426 and #7427 in 5 minutes#
08:33 ShadowBot https://github.com/minetest/minetest/issues/7426 -- Fix narrow/utf8 difference in incoming/outcoming messages by numberZero
08:33 ShadowBot https://github.com/minetest/minetest/issues/7427 -- Fix typo in `builtin/game/register.lua` by numberZero
09:00 proller joined #minetest-dev
09:16 Player-2 joined #minetest-dev
09:22 thePalindrome joined #minetest-dev
09:37 Fixer joined #minetest-dev
09:58 ensonic joined #minetest-dev
10:02 nerzhul stujones11: i updated my NDK to v17 we will see
11:05 Krock joined #minetest-dev
11:14 ensonic joined #minetest-dev
11:16 Gael-de-Sailly joined #minetest-dev
11:27 ensonic joined #minetest-dev
11:47 ensonic joined #minetest-dev
12:00 longerstaff13-m joined #minetest-dev
12:10 Foz joined #minetest-dev
12:20 Krock FYI: Added the rather trivial character encoding conversion fix to #7424
12:20 ShadowBot https://github.com/minetest/minetest/issues/7424 -- Bugfix bugfix release 0.4.17.1 by SmallJoker
12:21 Krock however, it's not the plan to wait for any more bugfix but the one which makes the Android build to succeed (optimally without warnings)
12:23 sfan5 i thought the android thing was solved now?
12:23 Krock yet no reply from nerzhul how NDK v17 works
12:23 Krock *on how well
12:46 Sokomine joined #minetest-dev
12:59 entuland joined #minetest-dev
13:04 Gael-de-Sailly joined #minetest-dev
13:26 ensonic joined #minetest-dev
13:30 BillyS minetest.register_on_generated is called on each new mapblock, correct?
13:30 BillyS Whoops, wrong channel
13:38 stujones11 joined #minetest-dev
13:38 stujones11 nerzhul: GCC is no longer supported in ndk-r17
13:38 stujones11 https://developer.android.com/ndk/downloads/revision_history
13:39 stujones11 It won't actually be removed until r18 so maybe it still works, IDK
13:54 Player-2 joined #minetest-dev
14:06 nerzhul stujones, i prefer clang too
14:07 nerzhul how did you patched the code to use clang ?but currently i always got the same problem
14:07 nerzhul i prefer the cmake builds as it just works fine but it's many work for mt (i have an opened pr about it)
14:08 sfan5 ndk-build will use clang by default in r17
14:08 sfan5 unless you set NDK_TOOLCHAIN_VERSION, that forces gcc
14:10 stujones11 Yes, so it may or may not work.
14:11 stujones11 I can verify that r16 does work though and that should be good enough for 0.4.17
14:12 stujones11 r17+ should be fine for 0.5.0 but backporting those changes will be somewhat non-trivial
14:17 nerzhul sfan5 i don't find where it's set in our code
14:18 sfan5 then we should be using clang already
14:20 stujones11 sfan5: We are using clang now, its just has not been backported
14:20 nerzhul okay i got some fixes to push
14:20 sfan5 you mean for the deps? I'm talking about mt itself
14:21 stujones11 For everything
14:21 sfan5 what did you change to make mt itself use clang then?
14:23 stujones11 https://github.com/minetest/minetest/blob/master/build/android/Makefile#L39
14:23 stujones11 https://github.com/minetest/minetest/blob/master/build/android/jni/Application.mk#L4
14:24 nerzhul i'm on clang now
14:24 nerzhul or not...
14:24 nerzhul it's gcc ld
14:24 nerzhul yes it's compiled by clang but linked by gcc ld
14:24 sfan5 https://github.com/minetest/minetest/pull/6628/files#diff-7e7f72f07eb9c2fe38f1c9d52534e829R4
14:24 sfan5 you added this line yourself
14:24 sfan5 so it would have defaulted to clang anyway
14:25 nerzhul no :)
14:25 stujones11 The build I made yesterday definitely used gcc
14:25 nerzhul i always have the linking issue, but now switched to the proper clang it's refused :)
14:26 nerzhul /home/nerzhul/Devel/Android/android-sdk/ndk-bundle/toolchains/arm-linux-androideabi-4.9/prebuilt/linux-x86_64/lib/gcc/arm-linux-androideabi/4.9.x/../../../../arm-linux-androideabi/bin/ld: warning: skipping incompatible /usr/lib/libatomic.so while searching for atomic
14:26 nerzhul /home/nerzhul/Devel/Android/android-sdk/ndk-bundle/toolchains/arm-linux-androideabi-4.9/prebuilt/linux-x86_64/lib/gcc/arm-linux-androideabi/4.9.x/../../../../arm-linux-androideabi/bin/ld: error: treating warnings as errors
14:26 nerzhul clang++: error: linker command failed with exit code 1 (use -v to see invocation)
14:26 sfan5 stujones11: 0.4.17?
14:27 stujones11 Yes, that is what this discussion is about
14:27 sfan5 yes, and what I'm saying is 0.4.17 would be using clang right now if you hadn't added "NDK_TOOLCHAIN_VERSION = 4.9" in your first PR
14:27 rubenwardy Clang is recommended on Android anyway
14:28 stujones11 nerzhul: Unless you backport the clang changes you will have to use gcc for this release
14:29 sfan5 gcc vs. clang shouldn't be the problem here anyway
14:29 sfan5 ld is searching in /usr/lib, why is it doing that?
14:31 nerzhul sfan5 i'm working on some build fixes for stable currently
14:31 nerzhul if you help me to fix this linking issue it should pass
14:33 nerzhul please note i forked libogg android here: https://gitlab.com/minetest/libvorbis-libogg-android
14:35 nerzhul also CEGLManager.patch has ending lines issues
14:35 sfan5 why are we even using that instead of using the normal libogg by running configure & make
14:35 nerzhul i don't know why we are using that, does your fork uses the regular ?
14:36 sfan5 I don't have a fork, but I would have used the regular libogg if I had written the build scripts
14:37 nerzhul i think so, but i just wanted to build it with current build
14:37 nerzhul we can change that in my refactor PR if needed
14:48 sfan5 every time i have to deal with our android build system, i feel like rewriting it
14:48 nerzhul https://github.com/minetest/minetest/pull/7123
14:48 nerzhul this is the solution, but it's very long to have a proper NDK + gradle build as many deps doesn't have it
14:49 nerzhul because target should have cmake too :(
14:51 nerzhul who disabled rtti in our build... erf :)
14:52 nerzhul it seems i fixed the build issue
14:53 sfan5 stujones11: do you have core.autocrlf in your git settings?
14:54 nerzhul it seems somewhere it forget to enable rtti i clean my whole workspace & rebuild
14:55 nerzhul i hope it will work after that
14:55 nerzhul we have some dynamic_cast in code else i should use C style old casts for this release :)
14:56 stujones11 sfan5: doesn't appear to be looking at `git config --list`
14:57 sfan5 hm
14:57 nerzhul not a big problem i can fix it
14:57 sfan5 something must be breaking the newlines of the patches
14:57 stujones11 You mean for #6750, right?
14:57 nerzhul it's the only failing patch
14:57 ShadowBot https://github.com/minetest/minetest/issues/6750 -- Android: Irrlicht patch has wrong line endings
14:57 sfan5 the thing with the patches is, the patch file itself is \n but the lines from the source file are \r\n
14:58 sfan5 in patch file in the repo is \n-only, which doesn't work
14:58 sfan5 stujones11: yes
14:58 sfan5 > Unfortunately this issue still remains even though I fixed these line-endings in 22a891a
14:58 sfan5 that depends on how the commit was merged
14:58 Gael-de-Sailly joined #minetest-dev
14:58 sfan5 git am discards (yes) all crlf
14:59 sfan5 if you try to transfer a patch that fixes the crlf over git am, it will end up empty
14:59 nerzhul i have the correct patch now :)
14:59 sfan5 see https://git-scm.com/docs/git-am#git-am---no-keep-cr
15:00 nerzhul when my build issues are fixed i will push fixed to stable-0.4 and backport needed changed to master
15:00 nerzhul i hope MT will build now :)
15:01 nerzhul and... no :) rtti remains disabled wtf
15:01 nerzhul pushing the current changes to stable to fix deps builds
15:02 nerzhul see https://github.com/minetest/minetest/commit/0ef9c53a8c20872ce69f1c539167497a04efe529
15:03 sfan5 nerzhul: i can reproduce the libatomic.so error here
15:03 nerzhul okay, switching from gnustl to c++_static disabled the rtti feature, i need to find how to solve it
15:03 nerzhul sfan5 take my last commit ^
15:04 nerzhul for RTTI: APP_CPPFLAGS := -frtti
15:04 nerzhul i will publish the patch when MT build will be okay
15:05 sfan5 LOCAL_DISABLE_FATAL_LINKER_WARNINGS=true works for me fyi
15:05 sfan5 do we really need a fork of the ogg vorbis repo?
15:05 nerzhul sfan5 i tried it but it doesn't work for me
15:05 nerzhul i updated the lib to use more recent toolchain to avoid warnings
15:05 nerzhul and it's under our namespace instead of a random guy, it ensure we never loose it
15:06 sfan5 https://0x0.st/s_Gc.txt
15:06 nerzhul for me the fix was the switch from ndk-build to make-build-toolchain like other deps :p
15:07 nerzhul (more proper than ignoring error anyway)
15:07 nerzhul i got the apk now, nice
15:07 nerzhul i need to update my local minetest_game and repackage it
15:07 nerzhul i pushed the rtti fix in stable 0.4
15:07 nerzhul we now have a make release working properly
15:08 sfan5 the patch i posted is enough to get an apk
15:08 sfan5 no need to touch anything rtti related
15:09 nerzhul you have more chance like me :)
15:09 nerzhul did you see this intersting shit https://github.com/minetest/minetest/commit/0ef9c53a8c20872ce69f1c539167497a04efe529#diff-75fe509a951ccb40a0e75444c93debaaL7
15:09 nerzhul there were a typo removing the C flags on our builds since a long time :p
15:10 sfan5 huh
15:10 nerzhul oh
15:10 nerzhul it's stujones
15:10 nerzhul we have same thing on Irrlicht.mk
15:10 nerzhul i push the fix
15:11 nerzhul b5350e27ad7bc54a3a97db988119db83c67f96b3
15:11 nerzhul nota long time, then, just added after 0.4.17 :)
15:13 nerzhul okay, my mtg is on 0.4.17 tag, i now rebuild the whole workspace with last patches and publish the APK on GH and play store after testing it on my mobile
15:13 stujones11 nerzhul: that was fine for gcc :p
15:14 nerzhul yeah, but it's deprecated, and now we are just up to date with clang
15:14 nerzhul next step: move to cmake + gradle haha (hard)
15:14 stujones11 Have the line endings also been fixed in master?
15:15 nerzhul not in master
15:15 nerzhul i'm on stable-0.4 currently
15:15 sfan5 this means we can release 0.4.17.1 now, correct?
15:15 nerzhul if sfan5 is okay with my commits i will push them to master
15:15 stujones11 Then you should not have closed my issue
15:15 nerzhul (after relasing)
15:15 nerzhul stujones11 we will upport them to master
15:16 nerzhul sfan5, we have more commits to add to the branch ?
15:16 sfan5 nerzhul: please push the line endings fix to master, not sure about the rtti things (really needed?)
15:16 sfan5 either way is fine with me
15:16 sfan5 nerzhul: yes see #7424
15:16 stujones11 okay, just don't forget :)
15:16 ShadowBot https://github.com/minetest/minetest/issues/7424 -- Bugfix bugfix release 0.4.17.1 by SmallJoker
15:17 nerzhul stujones11 after i'm able to perform a full 0.4.17 release yes i backport
15:17 nerzhul sfan5, i'm okay for waiting your bugfix release, push your changes to stable-0.4
15:18 nerzhul stujones11: i confirm the build now pass like a charm without any manual changes (after my mods)
15:18 nerzhul the only point problematic is the .1, android version is 0.4.17.<build version> already
15:20 stujones11 nerzhul: Did you test the apk yet?
15:20 stujones11 There is another annoying issue that can be easily fixed
15:21 sfan5 nerzhul: there are no more commits needed for a working android build, correct?
15:21 nerzhul stujones11: i just signed the apk, i will publish it to my phone now
15:21 sfan5 (did you bump the version?)
15:21 Lone-Star joined #minetest-dev
15:21 nerzhul sfan5: fb4bfc60deeab6c54e9c51f7f7a0ccf595e42b77 is my last fix commit and it's on the stable-0.4
15:21 sfan5 I know, just making sure
15:22 nerzhul i don't bumped any version
15:22 stujones11 https://github.com/minetest/minetest/issues/7409#issuecomment-395986898
15:22 nerzhul stujones11 let me test it :)
15:22 nerzhul app crash directly yes
15:23 stujones11 No crash, just a warning for me
15:23 sfan5 well looks like we need to push some more fixes to stable-0.4
15:23 stujones11 Newer versions of android may refuse to run it, IDK
15:24 nerzhul sfan5: version is fine, 0.4.17.18 (the 18 was not increased, i should do it when the release is okay)
15:24 sfan5 well yes but you have to increase versionCode to publish it to the store, no?
15:25 nerzhul yes, i will do it when i got the right apk :)
15:28 sfan5 ok
15:29 nerzhul version is okay, the next code should be 18
15:29 Krock the current code is 18 (0.4.17)
15:29 sfan5 it was already bumped once though
15:29 nerzhul i perform a debug build
15:30 Krock but if there was no android build, it probably won't need another bump
15:30 nerzhul the release is not sufficient to define why we crash, logcat is empty, it just said there is a problem
15:30 nerzhul Krock: generally i bump version after releasing on play store
15:31 Krock ok
15:36 nerzhul erf
15:36 nerzhul 06-10 17:35:43.240 27636-27649/net.minetest.minetest E/HAL: load: id=gralloc != hmi->id=gralloc
15:36 nerzhul it's the only error
15:43 stujones11 nerzhul: For clang you likely need to use APP_STL := c++_shared
15:43 nerzhul both should work, but maybe it's the issue there
15:43 stujones11 Otherwise it crashes on start-up, iirc
15:44 nerzhul stujones11 i try it
15:44 nerzhul the current error is not very verbose and i remember a such think on my gradle + cmake branch
15:45 stujones11 you will need to update that in both Application.mk & Deps.mk
15:45 nerzhul yeah i remember :)
15:46 nerzhul yeah it's that
15:46 nerzhul testing now
15:48 nerzhul sorry Krock :)
15:49 nerzhul i pushed the last change
15:49 nerzhul i'm connecting to a random server which has a verification to see if a pseudo sounds to be a real speudo xD
15:49 nerzhul cdlsdle is not allowed
15:51 stujones11 Did you not get the unauthorized access to /system/lib/libssl.so warning?
15:51 stujones11 IMO That really needs to be fixed for release
15:54 longerstaff13-m joined #minetest-dev
16:03 nerzhul stujones11: no i didn't see it
16:03 nerzhul my device is android 6
16:03 nerzhul maybe it depend on android/hardware vendor
16:04 nerzhul https://github.com/minetest/minetest/releases
16:09 stujones11 Yes, I am using version 7.1.1, on newer versions it may even refuse to run
16:10 stujones11 The fix is very simple, see my comments on the ogg issue
16:11 stujones11 This is already done in master
16:15 nerzhul stujones11, if we remove the ssl library loadding serverlist will not work as we need SSL library to perform HTTPS calls, same for some servers with https enabled content
16:19 stujones11 that fix was actually suggested by basicer, iirc and is done in current master
16:19 stujones11 I do not recall there being any problem with the serverlist in 0.5.0-dev
16:20 stujones11 however, I have 0.4.17 installed now so no way of checking without a re-build :(
16:22 nerzhul apk published https://github.com/minetest/minetest/releases/tag/0.4.17.1
16:22 nerzhul i publish it on the play store, we will see if there is a problem, but they should already have it and i don't see any comment in that direction
16:23 stujones11 okay, we'll see then :)
16:24 nerzhul interesting, apk is smaller than 0.4.16
16:24 nerzhul 12.7 for 0.4.16 & 11.43 for 0.4.17.1. Maybe because we drop 5 android versions (9 to 13)
16:24 stujones11 probably 'cos clang
16:25 nerzhul if i remember we optimized media in mtg in this releases :)
16:25 stujones11 that would help too
16:26 YuGiOhJCJ joined #minetest-dev
16:26 nerzhul oh
16:26 nerzhul interesting playstore note
16:26 nerzhul API 28 is mandatory for new apps on aug 2018, and for updated apps on november 2018
16:27 nerzhul we currently have exactly 7500 installations
16:27 nerzhul apk published on the play store (0.4.17.1)
16:28 Krock great.
16:30 nerzhul and i bumped the version on both branches. Master has also a gradle version update
16:33 nerzhul please note i push a tiny fix to disable clang-tidy check on non C/C++ modified files (like the linter)
16:33 sfan5 website updated too
16:34 nerzhul nice sfan5, ty
16:40 sfan5 rubenwardy: https://github.com/minetest/minetest/blob/master/builtin/mainmenu/dlg_contentstore.lua#L23 wtf is this
16:41 sfan5 1) you create a folder on startup that is never deleted
16:41 sfan5 2) this crashes on windows
16:49 sfan5 greeat
16:49 sfan5 this is what I get for not testing PRs myself
16:49 Cornelia joined #minetest-dev
17:15 Krock wasn't the os.tempfolder issue fixed recently?
17:16 rubenwardy Crashing on windows isn't my fault
17:16 rubenwardy And folders inside there are deleted
17:17 rubenwardy Oh no, screenshots aren't
17:17 rubenwardy Nevermind
17:17 rubenwardy This was a problem with the old mod store tio
17:17 rubenwardy I took that code from there, essentially
17:18 sfan5 Krock: the exact opposite
17:18 Krock well great. should've better downloaded or compiled a Windows binary to ensure that stuff works on _all_ Windows setups
17:19 sfan5 probably a mingw vs msvc thing
17:46 elinor joined #minetest-dev
18:17 Cornelia joined #minetest-dev
18:36 ensonic joined #minetest-dev
18:43 Darcidride joined #minetest-dev
19:21 nerzhul FYI the clang-tidy on non C/C++ commits is working properly :)
19:32 Gael-de-Sailly joined #minetest-dev
19:58 Calinou does clang-tidy operate on files other than C/C++?
20:29 joost_op_ joined #minetest-dev
20:32 joost_op_ Hi all. You guys. A while ago I've added Minetest to all Sabayon Linux images without asking any permission. I guess it is OK to ship our images with this software right? https://github.com/Sabayon/molecules/commit/c335a8944929575a9d334409ef2461806a665064
20:32 joost_op_ Just asking here to be sure.
20:35 troller joined #minetest-dev
20:36 nerzhul Calinou: no
20:36 Miner_48er joined #minetest-dev
20:39 Calinou joost_op_: should be OK, no license terms forbid you from doing so :)
20:41 troller joined #minetest-dev
20:42 joost_op_ Thanks Calinou, if there is some issue with it anyaway rightfull owners feel free to contact us. This game makes a great addition to our distribution live images.
20:42 joost_op_ thanks!
20:48 Lymkwi joined #minetest-dev
20:48 joost_op joined #minetest-dev
21:32 Darcidride joined #minetest-dev
21:52 troller joined #minetest-dev
22:08 BillyS Hmm, I got an interesting idea
22:08 BillyS A command like /deleteblocks that would respect modifications
22:09 BillyS So it would store the modified nodes, delete the mapblocks, and then re-add the modified nodes
22:09 BillyS Good for when a new ore is added or something like that
22:47 stujones11 left #minetest-dev
23:53 srifqi joined #minetest-dev
23:55 srifqi nerzhul: Are you sure you have tested to play before publishing the APK? game#2144
23:55 ShadowBot https://github.com/minetest/minetest_game/issues/2144 -- Error in init.lua while loading fireflies mod on minetest android build

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