Minetest logo

IRC log for #minetest-dev, 2018-04-12

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

All times shown according to UTC.

Time Nick Message
00:16 Zigara joined #minetest-dev
00:28 Cornelia joined #minetest-dev
00:44 octacian_ joined #minetest-dev
00:46 red-001 joined #minetest-dev
00:58 ssieb joined #minetest-dev
00:59 paramat #7234
00:59 ShadowBot https://github.com/minetest/minetest/issues/7234 -- Clientside FOV: Raise lower limit to avoid zoom-loading distant world by paramat
01:00 Cornelia joined #minetest-dev
01:30 Tmanyo joined #minetest-dev
02:29 paramat joined #minetest-dev
02:36 AndroBuilder joined #minetest-dev
02:37 AndroBuilder__ joined #minetest-dev
02:38 AndroBuilder___ joined #minetest-dev
02:39 AndroBuilder____ joined #minetest-dev
02:42 AndroBuilder joined #minetest-dev
05:04 Darcidride joined #minetest-dev
06:36 Lone-Star joined #minetest-dev
08:01 lisac joined #minetest-dev
09:02 Raven262 joined #minetest-dev
09:48 red-001 joined #minetest-dev
09:56 Fixer joined #minetest-dev
11:06 troller joined #minetest-dev
11:15 Player-2 joined #minetest-dev
11:15 longerstaff13-m joined #minetest-dev
12:16 Krock joined #minetest-dev
12:19 Icedream joined #minetest-dev
12:23 DI3HARD139 joined #minetest-dev
12:48 Wuzzy joined #minetest-dev
12:50 srifqi joined #minetest-dev
12:59 longerstaff13-m_ joined #minetest-dev
13:24 antims joined #minetest-dev
13:35 longerstaff13-m joined #minetest-dev
13:42 proller joined #minetest-dev
13:57 longerstaff13-m joined #minetest-dev
13:59 srifqi nerzhul: I can't debug Minetest on my phone because this problem: https://issuetracker.google.com/issues/37093233
14:00 srifqi (About #7122)
14:00 ShadowBot https://github.com/minetest/minetest/issues/7122 -- Android: confirmation dialog is... unreable and crashes
14:00 srifqi I was going to debug and I then I found that I can't.
14:22 srifqi ... by using ndk-gdb I mean.
14:23 srifqi I think there is another way to find the cause.
14:35 octacian_ joined #minetest-dev
14:58 Natechip joined #minetest-dev
15:13 srifqi By adding android:debuggable=true, I can find the cause line-by-line.
15:14 srifqi lol, :d turns into emoji
15:37 twoelk joined #minetest-dev
15:41 Beton joined #minetest-dev
15:47 DI3HARD139 joined #minetest-dev
15:52 Gael-de-Sailly joined #minetest-dev
16:00 srifqi I've managed to fix SIGSEGV error. Now, I can't interact with the GUI.
16:46 proller joined #minetest-dev
16:55 Natechip_ joined #minetest-dev
17:16 octacian_ joined #minetest-dev
17:18 proller joined #minetest-dev
17:46 srifqi left #minetest-dev
17:55 JDCodeIt joined #minetest-dev
17:56 JDCodeIt Would this be the right place to discuss memory leaks in the code?
17:58 Krock yes.
18:00 JDCodeIt Perfect - I ran 0.5.0-dev under valgrind in Linux and found some questionable activity around ExtrusionMeshCache - the code eventually exits with about half a MB hung up in there
18:01 JDCodeIt It goes deep into Irrlicht, but my guess is the reference count is going up and not getting "drop() - ed" enough times
18:02 JDCodeIt Question for those who know the code well - do any worker threads use the WieldMeshSceneNode class?
18:04 ssieb joined #minetest-dev
18:13 paramat joined #minetest-dev
18:15 ssieb joined #minetest-dev
18:19 paramat about the suspected memory leak, see http://irc.minetest.net/minetest/2018-04-11#i_5277196 , concerning
18:20 paramat the post at 22:52
18:21 paramat why is memory use so high when returning to menu?
18:23 paramat game#2108 is a return of a previous PR but now with vertical blend
18:23 ShadowBot https://github.com/minetest/minetest_game/issues/2108 -- Biomes: Remove some warm and humid biomes at altitude by paramat
18:26 Krock paramat, happens when something's not deleted/freed after leaving the in-game state
18:27 paramat yes, from the size it perhaps can only be meshes?
18:28 paramat just a random guess
18:29 paramat #7234 is fairly trivial will merge in 3 hours if no objections
18:29 ShadowBot https://github.com/minetest/minetest/issues/7234 -- Clientside FOV: Raise lower limit to avoid zoom-loading distant world by paramat
18:29 Krock could also be the texture cache or other media that was loaded and forgotten
18:42 JDCodeIt Anybody know the answer to the threading question? I'd be concerned with threads accessing the global variable, non-thread-safely. ExtrusionMeshCache sets one, and tries to do it only once.
18:45 JDCodeIt Seems to happen during "the_game" object create. Then when that goes out of scope, not everything is getting freed - of course just about everything happens in there.
18:45 paramat will update 7234 good point
18:48 paramat celeron55 see logs of this channel from 18:00 onwards, can you help with this memory leak issue?
18:49 paramat JDCodeIt are you able to open a Github issue on this? otherwise these bits of information get lost on IRC
18:50 JDCodeIt Sure, will do. At this point I only have some empirical results - it happens, but not much else to go one. Following the bread crumbs.
18:50 paramat ok :)
18:51 Krock ensure the animals don't eat them away before you found the way
18:51 JDCodeIt Yeah, there is that whole moving target thing
19:08 proller joined #minetest-dev
19:18 JDCodeIt Engine - Issue #7235 posted
19:18 ShadowBot https://github.com/minetest/minetest/issues/7235 -- Memory is not freed when exiting a world back to the main menu
19:30 Player-2 joined #minetest-dev
19:53 paramat thanks
20:00 Fixer_ joined #minetest-dev
20:02 longerstaff13-m joined #minetest-dev
20:18 JDCodeIt Is the preferred way to propose code changes - fork, modify, pull-request?
20:19 nerzhul yes
20:20 JDCodeIt OK, I found a small part that can help with #7235
20:20 ShadowBot https://github.com/minetest/minetest/issues/7235 -- Memory is not freed when exiting a world back to the main menu
20:25 JDCodeIt At what point can I retire my fork so it doesn't get so cluttered.
20:26 Icedream joined #minetest-dev
20:28 Krock for trivial spelling/1-liner fixes, pastebin is fine too
20:28 Krock or pastebin-alike services
20:49 Krock JDCodeIt, *fork, create branch, modify, PR
20:49 Krock you'll have to clean your master branch after the merge to not mess up everything
20:50 JDCodeIt Should I use a broom or vacuum cleaner? :) You mean delete it or somehow bring it back together with minetest-master?
20:52 JDCodeIt OK, I think I already messed it up - I didn't create a new branch in my fork, just committed it to my master.
20:52 Krock you have three choices: 1) delete master & fetch from upstream (minetest/minetest), 2) git reset --hard HEAD~1  to remove the additional commit and prevent rebase issues, 3) use another branch, like "master2" and try to link it with upstream master
20:53 Krock ^ but all of this doesn't matter as long you only have one PR simultaneously
20:54 JDCodeIt Once that PR is comitted, then I can do antoher, or in the mean-time can I also branch if I find other fixes?
20:55 Krock in the meantime you can also create other branches and ensure the correct commits are included
20:56 Krock a dirty "base branch" for all new branches will make your life unnecessary hard - just that I said it once
21:01 JDCodeIt OK, so you recommend after this PR is committed to minetest/minetest I do 1) above, then use branches for any change proposals. Not quite sure in the web interface how to do that (delete my copy, then fork again?)
21:03 sofar is your master polluted?
21:04 Raven262 joined #minetest-dev
21:05 sofar then $ git checkout origin/master && git branch -D master && git branch master && git checkout master
21:05 sofar I think
21:05 sofar will reset it back to a proper upstream
21:05 JDCodeIt I deleted 6 lines in one file and committed to my master - so I guess so
21:05 sofar you can also just
21:05 sofar git reset --hard HEAD~1
21:05 sofar to remove the last commit
21:05 sofar (you'd lose it, obviously)
21:06 JDCodeIt Then I already created the PR on minetest/minetest
21:06 sofar if that gets merged soon, just wait for that first
21:07 JDCodeIt ...that sounds like a plan - so simple a change should not be a biggie.
21:09 Icedream joined #minetest-dev
21:18 paramat #7234 updated
21:18 ShadowBot https://github.com/minetest/minetest/issues/7234 -- FOV: Raise lower limit to avoid zoom-loading distant world by paramat
21:22 paramat JDCodeIt Github has a 'boot camp' in the help section that clearly guides you through the whole process of contributing
21:29 Gael-de-Sailly joined #minetest-dev
21:39 Krock paramat, can't we check whether the player may zoom and limit the "full_d_max" and "d_opt" vars afterwards?
21:48 DI3HARD139 joined #minetest-dev
21:48 paramat hmm erm
21:51 paramat we could, but the PR as it is will set the correct zoom fov before using 'adjustdist' so seems better as it is
22:09 Fixer joined #minetest-dev
22:13 paramat joined #minetest-dev
22:15 longerstaff13-m joined #minetest-dev
22:22 troller joined #minetest-dev
22:30 paramat i'll work more on 7234 it needs more testing
23:37 Taoki joined #minetest-dev
23:53 paramat #7237 Shara
23:53 ShadowBot https://github.com/minetest/minetest/issues/7237 -- Reconsider confirmation on new player registration

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