Minetest logo

IRC log for #minetest-dev, 2012-11-25

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

All times shown according to UTC.

Time Nick Message
00:24 SpeedProg joined #minetest-dev
00:39 Taoki joined #minetest-dev
01:36 hmmmm joined #minetest-dev
02:12 VanessaE_[L] joined #minetest-dev
04:03 hmmmm joined #minetest-dev
06:20 VanessaE joined #minetest-dev
06:20 hmmmm joined #minetest-dev
06:20 Taoki|away joined #minetest-dev
06:20 sstrandberg joined #minetest-dev
06:20 celeron55 joined #minetest-dev
06:20 nyuszika7h joined #minetest-dev
06:20 thexyz joined #minetest-dev
06:20 darkrose joined #minetest-dev
06:20 ecube joined #minetest-dev
06:20 sfan5|OFF joined #minetest-dev
06:20 toabi joined #minetest-dev
06:21 rsiska joined #minetest-dev
07:21 VanessaE joined #minetest-dev
07:27 VanessaE joined #minetest-dev
07:33 VanessaE there.
08:19 hmmmm where?
08:27 VanessaE right there, of course ----->
08:28 jin_xi joined #minetest-dev
09:28 Calinou joined #minetest-dev
09:36 PilzAdam joined #minetest-dev
10:23 saschaheylik joined #minetest-dev
10:32 PilzAdam celeron55, darkrose, thexyz, I suggest to merge the models branch from Taoki
10:59 darkrose is that even able to be cleanly merged in? seems to be based on an older codebase
11:00 Calinou tried it, it's ok, but it's not mature enough. too many bugs
11:00 Calinou (eg. model lighting makes everyone darker is someone is inside a block)
11:02 celeron55 i get a segfault
11:02 celeron55 inside irrlicht
11:03 celeron55 http://paste.dy.fi/PW1
11:03 celeron55 8)))
11:06 celeron55 oh, works on third run
11:07 celeron55 the other player is playing completely random animations
11:09 celeron55 now it seems to stand still
11:12 celeron55 these random animations are very odd
11:13 celeron55 if a new player joins the server and other don't move, they appear to be doing completely random things
11:14 celeron55 they stop it when they have moved
11:14 celeron55 actually it isn't even random, it just cycles through everything
11:14 celeron55 maybe it's an intended test, whatever
11:18 celeron55 these random segfaults aren't too impressive
11:20 celeron55 doesn't anyone else get them?
11:20 PilzAdam I have no problems
11:20 celeron55 we'd need a trace of that segfault with a version of irrlicht with debug symbols
11:20 celeron55 i'm too lazy to do that
11:21 celeron55 (because there is no debug version of irrlicht in arch's repo)
11:21 celeron55 debian would have that i guess
11:21 kotolegokot joined #minetest-dev
11:38 kotolegokot left #minetest-dev
11:43 PilzAdam celeron55, do the segfaults happen when joining with 2 clients to a local server?
11:43 celeron55 that is the test case i used
11:43 PilzAdam then i noticed that too
11:43 Taoki celeron55: My models code is likely not the reason for these faults, I looked at the code well and IIRC fixed all of them
11:43 PilzAdam the second client crashed randomly on login
11:43 celeron55 it is the smallest case where any models get used
11:44 Taoki As for random animations, it's very weird someone else is seeing them. I get that too, but I again checked the code and am sure they shouldn't be possible
11:44 Taoki Its normally a local issue
11:44 celeron55 anyway, that segfault-inside-irrlicht has to be fixed
11:44 BambooVase joined #minetest-dev
11:45 Taoki I don't think its from my models code, it was likely already there. I checked the code and couldn't imagine where it could happen
11:45 celeron55 after that, i think it's good for merging
11:45 celeron55 it doesn't matter if it's your model code or not :P
11:46 celeron55 the only thing that matters is that it exists
11:46 Taoki Yeah, it does have to be fixed overall
11:46 Taoki The random animations bug is also a problem because the code shouldn't be causing it to happen. I checked and there's no way it could miss the animation setup. Don't know what it could be...
11:47 Taoki Since even if animations are set before the mesh is available, updateAnimations is ran after. Or if it's set after, updateAnimations already set the animation fields which took care of that
11:48 BambooVase I have a question about the missing texture on inventory bug. If I edit the /usr/local/games/minetest_game/mods/default/init.lua and add inventory_image = "pic.here" , it will work but when I join online it doest work. So is there another init.lua for online or its stored on the server?
11:48 PilzAdam Taoki, why is this not 1? https://github.com/MirceaKitsune/minetest_game/blob/models/mods/default/player.lua#L56
11:49 Taoki PilzAdam: That's only a default when the mesh is created. It it not the cause, I tried setting animations there as well
11:49 Taoki The code below sets animations immediately by itself
11:49 jin_xi BambooVase: its serverside
11:49 Taoki It has to be 0, so the player_step function sees that and sets it to the appropriate animation (in that case 1)
11:50 Taoki Which happens next frame at latest
11:50 BambooVase jin_xi: so theres no fix? i tried the atlas=true/false and it doesnt work
11:50 Taoki PilzAdam, celeron55: Anyway, I did try a short debug on the issue with the random animations. I have an idea of where it happens, although it makes no sense at all.
11:51 Taoki In my branch, see content_cao.cpp line 1388, which is if(m_animated_meshnode == NULL)
11:51 Taoki That makes sure there's an animated mesh node before animating, otherwise a segmentation thingie would occur
11:52 Taoki However, if the player is set to mesh, it's impossible to never reach that part with a mesh node being available at client initialization time
11:53 Taoki Since updateAnimation() is ran twice: Either when an animation packet is received by the client, either after the entity's visuals are set (eg: after it's created). So each circumstance would catch the player entity after its animated mesh node was created
11:54 celeron55 hmm... well, that is a minor problem
11:54 celeron55 unless it is related to the crash
11:54 Taoki no, totally unrelated. Just the default animation problem
11:55 PilzAdam bbl
11:55 celeron55 i ran this under valgrind, and don't see any obvious new problems... and this time it happened to work (it takes many, many minutes to get into game)
11:55 Taoki I'd be surprised if that crash has to do with my models code, nothing I can think of could be causing it
11:55 celeron55 i'll try a second time if it'd segfault
11:56 celeron55 well, it could be a bug in irrlicht that you trigger somehow
11:57 Taoki Only thing the default player script does now is setting mesh and texture and animation. Those are well coordinated, and everything should be checked. I'll keep poking around the code just in case
11:59 celeron55 this is something related to textures
12:00 Taoki good to know. Maybe it is something from my code then..
12:00 celeron55 i don't quite understand what this irrlicht code does though
12:00 celeron55 but can you reproduce the segfault yourself?
12:01 BambooVase left #minetest-dev
12:03 Taoki celeron55: It happens here too, but I thought it's from existing code in master or something local. I'll try to look into it
12:03 celeron55 you should get a version of irrlicht with debug symbols or build it yourself
12:04 celeron55 if you have debian or ubuntu, just uninstall libirrlicht and install libirrlicht1.7-dbg or something like that
12:05 celeron55 and then either enable core dumps or run in gdb
12:05 celeron55 that way you will get something much more informative than this http://paste.dy.fi/PW1
12:05 celeron55 (that is, line numbers and some variables from inside irrlicht)
12:06 Taoki I'll try to find it, since I know where textures are set for this in my code
12:06 Taoki BRB
12:07 celeron55 there are two things in that topmost function that could cause the segfault
12:07 celeron55 either CurrentTexture[stage] points to a higher index in CurrentTexture than the size of it is, or texture is an invalid pointer
12:08 celeron55 (i recommend getting the irrlicht source code and poking around, it's quite simple stuff)
12:08 celeron55 or browsing around
12:10 celeron55 i am guessing there is some sequence of initializing things that minetest can use when joining to a server that causes some variable that ends up in there to be be uninitialized
12:11 celeron55 or something similar
12:12 celeron55 as for what valgrind told me, there are no accesses to uninitialized memory on succesful runs
12:13 celeron55 and it didn't segfault for 3-4 tries, which might mean something needs to happen quickly (if so, valgrind can never cause it because it's slow as hell)
12:13 * celeron55 the guess machine
12:28 Taoki Currently I'm going to comment out suspicious code and see when and where that fixes the issue, since I know where to start
12:30 Taoki Already found the spot actually. Line 1247 in content_cao.cpp. Commenting that out causes the crashes to go away... so need to see what in that function was possibly doing it
12:34 Taoki Hmm... yes. I think it's because I need to check if a material exists first before applying a texture to it. I wonder if Irrlicht has a function for that
12:37 celeron55 getMaterialCount()?
12:37 celeron55 the line you said doesn't contain anything useful in what you have in github, but i am guessing 8)
12:37 Taoki Oh yes, I think that should do
12:37 Taoki Yes not that line, but the functions under it
12:38 Taoki The function at its beginning I mean
12:40 Taoki Ahhh, I think I gust fixed it :)
12:41 Taoki Needed to return in the loop if it exceeds material count. Strange since there was only one texture, but oh well
12:41 celeron55 that means sometimes there are zero materials
12:42 celeron55 hmm
12:44 celeron55 kind of odd, but whatever
12:45 Taoki celeron55: If you can now, please update the code git and confirm that segmentation fault is gone, so we can be sure
12:45 Taoki For me it is, I re-logged in several times
12:48 celeron55 nope, not gone
12:49 Taoki odd. For me it is
12:49 Taoki Try something else please,
12:50 Taoki Comment out the entire content of the for loop at line 1215
12:50 Taoki for (u32 i = 0; i < m_prop.textures.size() && i <= m_animated_meshnode->getMaterialCount(); ++i)
12:50 Taoki maybe even the loop itself at first
12:50 jin_xi still crashes for me with latest git
12:51 celeron55 and no, i wouldn't call this odd at all 8) it probably depends timing or unitialized memory, so the interval of it occurring can vary wildly
12:52 celeron55 +on
12:52 Taoki Also try commenting out only the lines in that loop that use Material as i (the loop iterator, like getMaterial(i) )
12:52 Taoki jin_xi too if you can look into that now
12:53 jin_xi trying now with that block commented out
12:54 jin_xi now 2nd player can login without crash
12:54 celeron55 clearing out the whole material-setting code would appear to get rid of it
12:55 celeron55 wait, what
12:55 celeron55 i <= m_animated_meshnode->getMaterialCount()
12:55 celeron55 Taoki: are you sleeping again? 8)
12:55 Taoki celeron55: at's wrong with that?
12:55 Taoki **what
12:55 celeron55 <=
12:56 celeron55 fixing that doesn't fix the problem though
12:56 Taoki Should I use < ? Won't it miss the last material then
12:56 Taoki ok
12:57 Taoki celeron55: Can you comment out only the lines that access material by i ?
12:57 Taoki like getMaterial(i) or anything of that sort
12:57 Taoki I think it's that but need to be sure
12:57 celeron55 well, that is more than likely, because it's like the only thing it is doing
12:58 celeron55 there's nothing left if you remove them
12:59 Taoki but the crash does away too right?
12:59 celeron55 of course
13:00 Taoki I wonder if I can use material = m_animated_meshnode->getMaterial(i) then if(material){...}
13:01 celeron55 most likely there are no problems the code, at least indexing problems
13:02 celeron55 for example CAnimatedMeshSceneNode::getMaterial(u32 i) goes like this:
13:02 celeron55 if (i >= Materials.size())
13:02 celeron55 return ISceneNode::getMaterial(i);
13:02 celeron55 return Materials[i];
13:02 celeron55 it can't fail
13:04 Taoki interesting... I wonder why then
13:04 celeron55 i think the problem is not that code
13:04 Taoki Though in my case, adding that extra check did fix the crashes
13:04 Taoki I think it was that code too, since for me it fixed it. But I'll try to optimize it too
13:05 celeron55 i suggest trying more times
13:05 Taoki I will
13:05 celeron55 because i am pretty sure it will fail (as it failed for jin_xi still too)
13:06 celeron55 so nobody is going to try with a debuggable irrlicht?
13:10 Taoki Trying a lot of times... not a single crashg any more
13:10 Taoki With another player already logged in of course
13:11 jin_xi ok, building irrlicht debug now
13:12 Taoki Thanks. I'm totally not good at things that deep so it's hard for me to debug it. Please update the code from my GIT again also, fixed the <= to < just in case
13:13 * jin_xi is not good at this
13:13 jin_xi trying anyway 8)
13:14 Taoki But yes, I tried a lot and the new change to check material count as well fixed it. Dunno how and why if others are still experiencing the crash
13:15 Taoki I'm running minetestserver in console and connecting two clients, all locally on 127.0.0.1
13:15 Taoki Keep logging one of them in and out and restarting the client but now it works
13:16 Taoki Tried more than 20 times, whereas previously the crash would happen half of the time
13:16 nyuszika7h joined #minetest-dev
13:17 jin_xi still had crash with latest git
13:17 jin_xi debug lib still building
13:17 jin_xi and its done. ok, now how i'm gonna make minetest use the debug version of irrlicht
13:18 Taoki jin_xi: I think you give it a different path to Irrlicht source and libs in cmake
13:18 celeron55 ok, i now have a segfault with a debug build of irrlicht 1.7.2
13:19 jin_xi aww shit i've done 1.7.3
13:19 Taoki Shouldn't be a big difference
13:20 Taoki I have my distribution's Irrlicht, I think that's 1.7.3 too
13:20 celeron55 well, i had the sources of 1.7.2 laying around
13:20 Taoki Nope, 1.7.9 here
13:20 celeron55 http://paste.dy.fi/Pwz
13:21 Taoki Maybe that's why, and 1.7.9 fixes this once I add the material count check
13:21 Taoki Can anyone try with Irrlicht 1.7.9 too?
13:21 Taoki 1.7.3 sounds so old either way :P
13:22 Taoki That might explain why that check does fix the crash for me
13:23 celeron55 umm... does irrlicht 1.7.9 even exist?
13:23 Taoki my repository shows that version of the Irrlicht package
13:23 celeron55 1.7.4 is the newest release in the 1.7 series
13:23 Taoki Sorry, then my distribution has weird versioning :P
13:23 Taoki I'll look again
13:24 celeron55 1.8 exists too
13:24 Taoki 1.7.9.3629-4.1 version of the irrlichtdevel package
13:24 Taoki odd
13:25 Taoki Hard to tell then, I likely have 1.7.3 or 1.7.4 too
13:27 celeron55 aanyway, it looks like the texture is somehow borked
13:27 celeron55 the failing thing is texture->getDriverType()
13:27 celeron55 which means texture points to crap
13:29 celeron55 it would be VERY useful to get this to fail under valgrind
13:30 celeron55 it might directly tell what is going on
13:35 PilzAdam i just builded latest git and it still crashes when the second client is logging in
13:36 Taoki PilzAdam: I'm the only one who no longer gets the crash oddly enough
13:37 Taoki Makre sure you have both my latest code branch and minetest_game branch to be safe. But yeah, it's weird
13:37 jin_xi i cant even get it built with 1.7.3 debug... im sure i must be doing sth wrong
13:37 PilzAdam i tried without minetest_game
13:37 Taoki Erm... without minetest_game my code can't cause the crash, since the lua script needs to set the players to mesh :P
13:38 Taoki If you get this without the player.lua script it cannot be my code, since that area is never reached
13:38 PilzAdam well, it crashed
13:39 PilzAdam btw: without minetest_game the player textures are wrong; random textures are taken (e.g. lava bucket)
13:40 Taoki PilzAdam: Unrelated to my code again, I think that was introduced with the new animated water or something
13:40 Taoki That's when it started happening for me too, and I get bad textures on the 3D player also
13:40 Taoki Maybe that has to do with it?
13:40 celeron55 i am trying to figure out where CAnimatedMeshSceneNode::render() is getting the material with screwed material.TextureLayer[0].Texture
13:40 celeron55 it can get it from two sources
13:40 Taoki celeron55: From what PilzAdam is saying, the existing crash might be to something other than my code. Since he gets it without the player script, which means that code isn't even reached
13:41 celeron55 it is either CAnimatedMeshSceneNode::Materials[0] or CAnimatedMeshSceneNode::getMeshForCurrentFrame()->getMeshBuffer(0)->getMaterial()
13:45 celeron55 hmm... very true
13:45 celeron55 in that case it happens in irr::scene::CMeshSceneNode::render
13:46 celeron55 looks similar, http://paste.dy.fi/oHB
13:47 celeron55 i don't think there are other options than bisecting
13:48 celeron55 i hope most of Taoki's commits will build, because otherwise this will be horrible; very horrible
13:48 Taoki Ah... it could be the anisotropic filtering lines, I added that for all entities (including sprite and upright_sprite)
13:48 Taoki Sure they will :)
13:48 Taoki I compile and test nearly each time before pushing to GIt
13:51 Taoki But yes, this is very weird and confusing then
13:53 celeron55 [celeron55@nk minetest]$ git bisect start HEAD 25cf3757b29a0397e201a4719788ead3dc76ec94
13:53 celeron55 Bisecting: 66 revisions left to test after this (roughly 6 steps)
13:53 celeron55 8)
13:53 celeron55 i'd like doing this more if i was the one who broke this
13:54 celeron55 and had a 100x faster computer
13:54 Taoki celeron55: So, should I worry about my models code any more? Since like I said I no longer get the crash, and PitzAdam gets it without the player.lua script even, which means the texture setting code doesn't get there
13:55 celeron55 it's definitely something you did
13:55 celeron55 there are really no other possibilities
13:58 Taoki odd then
13:58 Taoki celeron55: Please try one more thing if you can. Comment out all the lines I added in content_cao for bilinear trilinear and anisotropic filtering
13:59 Taoki Because I added that as well to all entities after Vanessa's patch, so filtering will work on 3D players
13:59 Taoki Maybe it's that, though I also have those lines enabled in my code and no crash
13:59 celeron55 i am doing a full bisect
14:00 Taoki ok. The easiest way I use personally is to comment out sections of code until I slowly get to the root of the issue, typiocally one or two lines. But checking GIT revs can help even more
14:00 celeron55 2434502af8d0338674ddeae80ddae75a3f4e237f doesn't build 8)
14:01 Taoki There was one revision that didn't build, I only remember one though :)
14:01 Taoki Since 95% of the time I compile then test in-game before committing
14:21 celeron55 41c00e87d42580881d8c7001f39014814e3b55f2 is the first bad commit
14:21 celeron55 commit 41c00e87d42580881d8c7001f39014814e3b55f2
14:21 celeron55 Author: Kahrl <kahrl@gmx.net>
14:21 celeron55 Date:   Thu Feb 2 04:11:56 2012 +0100
14:21 celeron55 Don't leak textures all over the place
14:21 celeron55 :040000 040000 9d202bb0a38e3a83d9814bc6cfd8bbcf72ff3316 468a3891255079c8b7d54e2a9c57d851b4d55b9f M      src
14:21 Taoki Ahh, so it was a commit unrelated to my code after all
14:22 Taoki If it can be fixed in master let me know and I'll update my branch
14:23 celeron55 this would mean that celeron55/master should segfault similarly
14:23 celeron55 i have never seen it do that
14:24 Taoki Maybe that only happens combined with my code, somehow. Or just coincidence it hasn't been noticed so far
14:25 PilzAdam current upstream segfaults
14:25 PilzAdam just tested
14:25 celeron55 lol
14:25 jin_xi lol, noone plays on a server using head
14:26 PilzAdam noone plays on local server with 2 local clients
14:26 Taoki Heh, and I might not be getting them because I haven't updated my branch with master in the last 5 days, or maybe it just doesn't happen here
14:26 Taoki But let me know when the problem is fixed and I'll merge new master into my models branch
14:26 celeron55 you do have merged that part of upstream into yours
14:27 Taoki PilzAdam: Still, why would playing on local server with 2 local clients make a difference? From connecting to another server with only one local client or anything else
14:27 Taoki celeron55: Probably I can switch and tell you when my local master has been updated (whcih is inside my models branch) just a moment
14:27 PilzAdam because nobody ever noticed that on not local servers
14:28 PilzAdam the commit was Nov 02, 2012
14:28 jin_xi nice combination of events
14:28 celeron55 https://github.com/MirceaKitsune/minetest/commits/models?page=3 "Don't leak textures all over the place"
14:28 jin_xi the patch is much older
14:28 Taoki celeron55: My latest local master is 3ba9b263103d2cd5218246b05d19b6485347ae28
14:28 Taoki Fix server crash on /clearpassword
14:28 Taoki According to #253, using `/clearpassword` without an argument causes the server to crash from an assertion failure. I've resubmitted matttpt's patch as a pull request to aid in merging.
14:29 Taoki If the problematic change was before that then yes, I have the commit
14:29 PilzAdam it was 18 days before
14:29 Taoki Author date 19 Nov 2012 01:18:45
14:29 Taoki yep, then I have it too
14:30 Taoki My models code that is
14:37 celeron55 this texture deletion thing needs to be rewritten to work in a completely different way for it to work
14:38 celeron55 i'll revert it
14:39 jin_xi it did not fully fix texture leak problems anyway
14:39 Taoki ok. will merge latest master in my code then it can be tested again. That check to nor exceed mesh material number was also needed, so I'm glad I got around to that
14:39 PilzAdam it fixes my texture problems
14:39 Taoki Since without it specifiny more materials / textures than the mesh had in the script would cause a fault by itself
14:39 Taoki **specifying
14:42 celeron55 pushed
14:44 celeron55 it wouldn't have crashed to that
14:45 celeron55 it would just have been modifying the video::IdentityMaterial
14:45 celeron55 which is kind of bad too though 8)
14:49 Taoki celeron55: Merged the fix in the models code, re-compiled and tested. Please update my GIT and see if the crash with my models ocrrus any more
14:49 Taoki If not then it's clearly fixed everywhere
14:50 rubenwardy joined #minetest-dev
14:52 Taoki jin_xi, PilzAdam: Please update my branch too if you're still at that, so we can be sure any crash is gone for good. Like I said I can no longer reproduce any crash here
14:52 jin_xi building it..
14:52 Taoki sure
15:02 jin_xi ok
15:02 jin_xi no crashes anymore
15:03 celeron55 it does seem to work
15:10 celeron55 somebody could experiment with calling video::IVideoDriver::removeAllTextures() at the end of the_game()
15:10 Taoki Sweet... means the problems are all gone in that case
15:11 Taoki Since I also fixed the random animations, by calling updateAnimations in a better place. So when a new client connects it will now see a player's proper animation looping, even if nothing else triggers a new animation since then
15:12 Taoki No idea why that was needed but it's correct and works, so yeah. Should re-test attachments just to be 100% sure
15:12 celeron55 hmm
15:13 Taoki I still need to discuss something in the script with RealBadAngel when he's back. After that, I might actually pull request the code and minetest_game branches tonight. Since after all this there are no major issues known any more
15:13 celeron55 maybe some of your commits should be squashed together
15:13 celeron55 there are so many of them and nobody really cares of the halfway things
15:13 Taoki yeah. Not good at doing that however, never have before in GIT.
15:14 jin_xi ok, testing attachment
15:14 Taoki thanks. The latest player script removed all those test cases since they were temporary
15:14 Taoki Only time we'll have a permanent attachment function for players there is once someone better at scripting will help implement wielded items
15:15 celeron55 Taoki: well, i can tell you how to do it
15:15 Taoki But that can come later, the scropt should make it easy
15:15 Taoki celeron55: With some very painful console commands? :P
15:15 celeron55 they're not painful
15:15 Taoki Or maybe github can help, though it wouldn't update the local branch
15:16 celeron55 first make a new branch to do the stuff in with "git checkout -b models_rebased" (that makes one called "models_rebased" and makes it active)
15:17 celeron55 then make sure your git uses an editor you are familiar with
15:17 Taoki I use qgit
15:17 Taoki Anyway, wouldn't it be possible to just delete some commits that are unimportant?
15:19 celeron55 okay well, let's do it an another way
15:20 jin_xi attachments seem to work as well
15:20 Taoki awesome
15:20 celeron55 this is what you get by calling "git rebase -i 25cf3757b29a0397e201a4719788ead3dc76ec94": https://gist.github.com/4143935
15:20 celeron55 they are from oldest on top to newest at bottom
15:20 Taoki Should... I just moved the updateAttachments function somewhere else which should make little difference
15:21 celeron55 the order can be modified by moving the lines up and down, and if you change the "pick" at the left to "s", the commit will meld together with the previous commit
15:22 Taoki ah, i can edit it on that page?
15:22 celeron55 git rebase -i gives this in an editor, that you can edit, and save, and then it'll modify the history according to it, and ask for new commit messages for those that were squashed or so
15:22 celeron55 if you edit it on that page, i can just copy it back to my editor and git'll do it
15:22 Taoki ok. Can I then just git push to update the github branch?
15:22 Taoki sure
15:22 Taoki That sounds ok
15:22 celeron55 don't remove any line, and don't duplicate any line (it will make things bad)
15:23 celeron55 and commit messages cannot be edited at this point
15:23 Taoki that's fine too
15:23 celeron55 (they will be asked later if one puts 'r' at the beginning of a line)
15:23 Taoki So how should I edit those? Just change "pick" to "s" on comments I'm ok with being mended?
15:23 celeron55 but basically... try to make that list to be 10 meaningful commits or so
15:24 celeron55 well
15:24 Taoki I'll make the commits I want gone with "s" then
15:25 celeron55 commits that can be squashed together with the previous one
15:25 rubenwardy1 joined #minetest-dev
15:25 Taoki sure... will get working on it. Then that should mean you can merge the code branch upstream :)
15:25 celeron55 so that they will make a meaningful one commit that fully makes one thing properly
15:26 celeron55 hmm, i'll do an example
15:27 celeron55 that list of commits is like 10x larger than anything i've edited before
15:27 celeron55 kahrl would have rebased everything in it to one commit and then given it to me :P
15:28 Taoki I'm editing it now, and adding s to all commits that can go. Will pastebin when ready
15:30 celeron55 basically, if there is stuff like "pick do thing X" "pick fix something in X" "pick do Y" "pick fix more in X", you can put them like
15:30 celeron55 "pick do thing X" "s fix something in X" "s fix more in X" "pick do Y"
15:31 celeron55 i don't think you can make those reasonably to 10 commits... but try to make it a bit shorter :P
15:32 Taoki I'll try to leaqve less than 15 or so commits with pic, and mark the rest with s
15:33 celeron55 actually.... uhm
15:34 celeron55 this will create problems
15:34 celeron55 don't do anything :D
15:34 Taoki Just marking them in a text editor, so I at least have a personal list of the commits that are ok to stay
15:35 celeron55 the reason is that it merges stuff from upstream... hmm, i am not sure
15:35 celeron55 well, we can try
15:35 celeron55 i'll try some small tests so i can see how this will break
15:40 celeron55 oh, yes
15:40 celeron55 this *will* work
15:40 celeron55 no problem at all 8)
15:40 celeron55 git is magic
15:43 Taoki I reduced it to 16 commits. Is that ok? Should be kept min ind that a lot of work was done both on attachments and models, which are sort of two different features that were done in the same branch because they were needed together
15:43 Taoki *kept in mind
15:43 celeron55 well, paste it somewhere
15:43 celeron55 if you want to be extra helpful, you could add the desired commit messages to there as comments :P
15:43 Taoki http://pastebin.com/rw06gubx or http://pastebin.com/raw.php?i=rw06gubx
15:44 Taoki I can add the commit lines only if that's easier
15:44 Taoki Like only paste the lines that begin with "pick"
15:46 celeron55 i think this will go as-is
15:46 celeron55 let's see 8)
15:48 Taoki If it's easier here is the liter version: http://pastebin.com/ScjE8h0F ---- http://pastebin.com/raw.php?i=ScjE8h0F
15:48 Taoki Well, there are about 19 commits, but they should be meaningful and ok
15:49 Taoki But yes, since you're working on that feel free to merge this up-stream already. The code is ready... only thing left are some tiny changes to the player script I'll discuss with RBA
15:49 Taoki Attachments were also tested locally by me a lot, then with the carts mod, and jin_xi just re-tested them again too... so no doubt there for me
15:50 Taoki Animations are being tested for 2 days since I added the player animation script and they seem to be working flawlessly now, especially as far as the code is concerned
15:50 Taoki Models and textures are set ever since I began the feature so those won't have the slightest issue
15:51 Taoki Sneaking also changes animation speed properly, so overall I can say animations were fully tested and should have no problem
15:55 saschaheylik joined #minetest-dev
15:56 rsiska joined #minetest-dev
15:59 * celeron55 feels like a machine when git slaps a bunch of 20 commits in the editor and asks him to figure out a reasonable commit message for them
15:59 celeron55 ...19 times
16:00 celeron55 i don't even know how many i've done already 8D
16:01 Taoki Don't the commit messages get inherited in this case? For the commits I am keeping and squashing to
16:01 celeron55 sometimes almost all text of all commits needs to be left, sometimes i can just erase everything but a few words of the main commit :P
16:01 celeron55 they get, but they're sometimes quite off what the whole bunch does
16:02 celeron55 then often there is stuff like "fix this" "fix that" "revert fix", and almost as often the important information comes in some of the squashed commits
16:03 Taoki joined #minetest-dev
16:04 Taoki back... KDE crashed the system again
16:08 celeron55 hmm, i have a bad feeling about this...
16:08 Taoki If you got the squashed version of my branch on a local branch, merging that to master should be no issue
16:09 celeron55 i'm worried about the inclusion of unrelated commits in these squashes
16:09 celeron55 it will hopefully work
16:09 Taoki Ah. I tried keeping the most important ones
16:09 * jin_xi crosses finges
16:09 Taoki Yes, hope it all works well
16:10 celeron55 this squashes stuff like VanessaE's filtering option GUI
16:10 celeron55 into some attachment commit 8)
16:10 rubenwardy left #minetest-dev
16:10 Taoki ahhh, that. Yeah
16:10 Taoki Prolly can't be helped
16:10 celeron55 git should be able to handle that reasonably, but it'd be better as a non-squashed one
16:15 celeron55 oh god, merge conflicts when rebasing the end result on master 8)
16:15 Taoki ah, those are always annoying >_<
16:15 Taoki Latest commit on my branch doesn't have merge conflicts BTW. But I imagine the squashed ones might
16:15 Taoki *with latest master
16:17 celeron55 hmm... done, kind of
16:17 celeron55 now building
16:18 Taoki awesome
16:18 celeron55 this might explode 8)
16:19 Taoki BTW. Is my anaglyph stereo branch in also? After this is over you could look into that too since I can be around for 30 more minutes or so
16:19 Taoki That's very small and has been ready and working for an year IIRC
16:20 celeron55 omg, it compiled
16:20 Taoki awesome :)
16:22 celeron55 it looks like this https://github.com/celeron55/minetest/commits/models_rebase2.master
16:23 Taoki Looks great
16:23 celeron55 not perfect, but way better than before
16:23 Taoki I assume the latest commit has the changes I made today right? Like checking for material count and moving updateAnimations somewhere else
16:23 Taoki yeah
16:24 celeron55 ugh
16:24 celeron55 RBA's keys-over-network patch got squashed into that last one
16:24 celeron55 oh
16:24 celeron55 it's only it
16:25 celeron55 well that is correct then
16:25 Taoki it's all good, those changes are in that commit now
16:25 Taoki Doesn't matter to me really... this is only if someone goes back months / years from now to see how things were done :)
16:26 celeron55 PilzAdam: can you think of any problem of putting this to master?
16:26 celeron55 oh, then there is the game part
16:27 Taoki yes. That can come later, but that one's ready too really. Only issue is the death animation looping constantly for dead player
16:27 PilzAdam putting what to master? (log is tl;dr)
16:27 Taoki PilzAdam: <celeron55> it looks like this https://github.com/celeron55/minetest/commits/models_rebase2.master
16:28 doserj joined #minetest-dev
16:28 celeron55 i think the minetest_game commits could be just squashed all into one
16:28 celeron55 they're all about fixing each other :P
16:28 Taoki celeron55: Would that little issue be acceptable for now, until me and RBA fix it? That when a player is dead, he will loop the death animation, meaning he constantly falls to the ground then gets back up and falls again etc.
16:29 Taoki I can see how to squash that one too, if we can do it the same way. I'd rather leave a few commits there
16:31 celeron55 that isn't a huge issue
16:31 Taoki ok. If it isn't then the rest should be all fine
16:31 PilzAdam celeron55, if the game doesnt crash with the new code merge it; it can be developed further in upstream
16:32 Taoki PilzAdam: From what we've been testing in the last hours, no more crashes of any sort with latest code
16:32 PilzAdam and +1 for put it into a single commit in _game
16:32 celeron55 Taoki: https://gist.github.com/4144257
16:32 Taoki I'll try to sqiash it to a few commits only, if celeron55 can paste another list
16:33 Taoki yes, that. Will get on it now :)
16:33 celeron55 try to leave all non-related commits intact this time
16:33 celeron55 like "Add animation for water"
16:33 sstrandberg joined #minetest-dev
16:34 celeron55 you can move them around a few commits so that you can squash the squashable stuff if you want
16:34 Taoki ok
16:35 celeron55 but eg. moving some of them to the beginning while not moving some others may cause problems :P
16:35 celeron55 (because the order changes)
16:35 celeron55 (order of editing one file)
16:35 celeron55 (or, well, the same portion of one file)
16:35 PilzAdam brb (10 mins.)
16:38 Taoki I will leave a few commits in, for changes I did to the player mesh as well as the script
16:40 Taoki celeron55: Done. http://pastebin.com/EMijgx90 ---- http://pastebin.com/raw.php?i=EMijgx90
16:42 Taoki Short version: http://pastebin.com/iBcxP0hY ---- http://pastebin.com/raw.php?i=iBcxP0hY
16:43 Taoki 10 commits by me, the others are what came from merging master into my branch and I left them like you said
16:45 PilzAdam_win joined #minetest-dev
16:46 Calinou joined #minetest-dev
16:47 hmmmm joined #minetest-dev
16:48 Taoki celeron55: I need to go in a minute, will be back in 1-2 hours. Is everything ok from here?
16:49 Taoki I'm ok with anything, just make sure the changes in the latest commit is in the latest commit of the resulting squash (sice I made some more important ones today to the player.lua script)
16:49 celeron55 there are some odd conflicts in rebasing this
16:49 celeron55 i need to guess how these test case changes need to go
16:49 Taoki ok
16:50 celeron55 i'm hoping they don't break, but this isn't exactly easy 8)
16:50 Taoki Hope it all works well. Need to run now, later (will be online as usual just afk)
16:56 kotolegokot_ joined #minetest-dev
16:57 celeron55 https://github.com/celeron55/minetest_game/commits/models_rebase2.master
17:00 celeron55 hmm, this has some difference to the current MirceaKitsune/master
17:00 celeron55 eh
17:00 celeron55 /models, i mean
17:00 Calinou looking at these commits, the lighting bug would still be present
17:01 Calinou if some player is in darkness or inside a block, ALL players will be dark or black. this happened this morning on RBA's server
17:01 Calinou when jordach was inside a block everyone was black
17:01 celeron55 the model would need to be copied for each object for the vertex lighting to work
17:01 celeron55 which can be done, and is done for items
17:02 celeron55 (wielditem meshes)
17:03 PilzAdam_win btw: talking about item entities: https://github.com/celeron55/minetest/pull/301
17:04 celeron55 wtf
17:05 celeron55 some patch in celeron55/models_rebase2.master reverts the reversion of the segfault commit
17:05 celeron55 8D
17:06 celeron55 lol
17:06 celeron55 this one https://github.com/celeron55/minetest/commit/996be396095833f5a9decb9e694c02bdb8cfe073
17:06 * celeron55 edits
17:12 Calinou using Taoki's models branch, when I make another player join my local server (using localhost), it segfaults
17:12 Calinou original player stays though
17:13 PilzAdam_win Calinou: thats not a bug in the models branch but in upstream (fixed now)
17:14 Calinou i'm recompiling
17:14 VanessaE odd that it didn't bug out on me for the several minutes I was playing with it last night on RBA's testing server.
17:14 Calinou nope
17:15 Calinou i'll see if i start a dedicated server
17:15 Calinou works
17:15 celeron55 this is now re-rebased: https://github.com/celeron55/minetest/commits/models_rebase2.master
17:16 sstrandberg joined #minetest-dev
17:17 celeron55 i'm waiting for Taoki to check https://github.com/celeron55/minetest_game/commits/models_rebase2.master , i am not sure if it's prefectly right
17:17 VanessaE celeron55: careful, you and taoki are going to drive PilzAdam crazy with such a high commit-to-push ratio ;)
17:22 darkrose joined #minetest-dev
17:22 darkrose joined #minetest-dev
17:24 celeron55 i think https://github.com/celeron55/minetest_game/commits/models_rebase2.master is missing some of the newest commits
17:24 celeron55 but i am not sure how many
17:25 celeron55 apparently three of them
17:25 celeron55 i'll add themn
17:25 celeron55 -n
17:25 PilzAdam joined #minetest-dev
17:26 celeron55 and done!
17:26 celeron55 how i want at least two people to test that end result
17:27 celeron55 these: https://github.com/celeron55/minetest/commits/models_rebase2.master https://github.com/celeron55/minetest_game/commits/models_rebase2.master
17:27 jin_xi ok
17:27 PilzAdam testing...
17:29 VanessaE testing...
17:29 saschaheylik joined #minetest-dev
17:29 celeron55 this should be the only bug: if some player is in darkness or inside a block, ALL players will be dark or black. this happened this morning on RBA's server
17:29 Calinou copy-pasta much
17:30 PilzAdam singleplayer... check
17:30 celeron55 and the reoccurring death animation
17:30 Calinou note that this does not always happen, and please test this with 3+ players
17:30 celeron55 it happens if a client sees two or more players
17:31 celeron55 then the lighting of some or the other is used for all of them
17:31 Calinou about the death anim, methinks it should stop at the least frame until player respawns. once a player dies, put the name tag in ()
17:31 Calinou (so that you know when a player is dead :p useful when people are AFK)
17:31 PilzAdam 2 clients at one local server.... check
17:32 Calinou you'll need 3, see what c55 said
17:32 VanessaE aside from having spawned at the bottom of a large, wet ravine, singleplayer seems okay here.
17:33 Calinou is it just me or since 0.4, player movement smoothing is broken? back when I started playing in 0.2, players moved very smoothly (not talking about turning, moving here)...
17:34 VanessaE 24.181.193.82 port 30001
17:35 VanessaE if you guys want somewhere to gather to try to break it.
17:36 VanessaE oops, forgot to adjust the spawn point.
17:36 jin_xi looks good to me
17:36 VanessaE wait.
17:37 VanessaE there, now you can connect.
17:39 VanessaE strange, this was fetched from the ...
17:39 VanessaE wait a sec.
17:39 PilzAdam you need the _game models branch
17:39 VanessaE odd, I thought I did that too.
17:39 VanessaE one sec.
17:40 VanessaE something went wrong.  Stand by while I figure it out.
17:40 Calinou k
17:40 * PilzAdam stands by
17:41 * Calinou shuts down
17:41 VanessaE there we go.
17:41 * celeron55 suspends
17:41 Calinou to RAM or to disk?
17:41 Calinou :D
17:41 VanessaE it's back, see if that works.
17:41 VanessaE yay!
17:41 VanessaE that's better.
17:42 VanessaE there's the everyone-is-black bug you mentioned, c55.
17:42 Calinou found a bug, if you have fly mode but no permission, you don't have a black screen when inside a block
17:42 Calinou :>
17:42 jin_xi because we're all stuck in a block at spawn
17:42 Calinou if you disable it then you have black screen
17:43 VanessaE jin, fly out.
17:43 VanessaE I'd forgotten to set the default spawn point.
17:44 * Calinou segfaulted
17:57 sstrandberg joined #minetest-dev
17:59 celeron55 oh dog
17:59 celeron55 this is going to take a day
17:59 VanessaE aw shit, it's using 2d sprites again.
17:59 celeron55 rule 1 in hosting minetest: never put shitloads of mods on a server that is supposed to be accessed from an another continent
18:00 PilzAdam yep
18:00 celeron55 loading at 25%
18:00 celeron55 41...
18:01 VanessaE wow, that's a while alright :-)
18:01 jin_xi lol i'll be there in 10 minutes ;)
18:01 celeron55 62
18:01 VanessaE odd thing is, I'm only putting out about 35 kB/sec.  I can do 300 kB/sec easy on this connection.
18:02 celeron55 minetest does congestion control based on round trip time
18:02 PilzAdam thats what the TCP branch is for
18:02 celeron55 so it sucks for big data transfers over long connections
18:02 PilzAdam 27%
18:03 VanessaE perhaps it's just too sensitive?
18:03 celeron55 no; if it would be tuned for a higher RTT, it would cog up low-RTT connections
18:03 VanessaE hm, good point
18:04 celeron55 it would need to be more clever, but a better thing'd be to just use tcp for bulk transfers, for which i made a branch, which has bugs
18:04 celeron55 81
18:04 VanessaE right.  well that'll get worked out eventually.
18:04 celeron55 i won't touch it
18:04 celeron55 it's horrible
18:04 celeron55 completely non-debuggable
18:04 VanessaE someone will come up with a solution I'm sure, if not that branch then something new.
18:05 celeron55 i'm waiting for some kind of a crazy person who'll rewrite the whole network stack or use his magic wand to find the jamming bug from the tcp_blocks branch 8)
18:05 VanessaE lol
18:05 PilzAdam 70%
18:05 VanessaE well you never know :-)
18:06 jin_xi lol 2d players?
18:06 VanessaE server going down while I fix what I broke :D
18:06 PilzAdam 79%
18:07 jin_xi pls lets just test carts mod with a handful of players
18:07 celeron55 kill all the mods
18:07 jin_xi as it uses the new features
18:07 celeron55 and put carts on it
18:08 VanessaE we're back.
18:08 PilzAdam 0%
18:08 VanessaE much better.  had an old copy of minetest_game embedded in the mods tree.  sorry about that.
18:09 celeron55 how many million mods did you put on it now
18:09 Calinou servers hould have ability to change max upload speed
18:10 Calinou "no; if it would be tuned for a higher RTT, it would cog up low-RTT connections" < that would make it possible, no?
18:10 celeron55 there is no such thing, it is dynamic
18:10 Calinou ah
18:10 VanessaE celeron55: no more than it had a minute ago, I actually deleted a few ;-)
18:10 celeron55 i'm still waiting for this thing to load
18:10 PilzAdam 70%
18:10 celeron55 i guess i'll set up a server myself
18:10 Calinou it'll lag even more :D
18:11 Calinou j/k
18:11 VanessaE celeron55: blame your network congestion code.
18:11 celeron55 no, i blame you for misuse of minetest
18:11 PilzAdam everyone except VanessaE are in europe?
18:11 VanessaE heh
18:11 PilzAdam and we let her host the server?
18:11 celeron55 let me set up a server in germany
18:12 celeron55 at the end of a huge network pipe
18:12 celeron55 = the minetest.net machine
18:12 VanessaE celeron55: I have a 3Mbps pipe here.
18:12 celeron55 it's probably 100M or more
18:12 VanessaE nice.
18:13 PilzAdam i had to exits because the huge amount of textures made my taskbar corrupted
18:13 celeron55 well, whatever, installing of that will take a while
18:13 VanessaE ok I'll cut the mods down to just carts then.
18:13 PilzAdam finally....
18:14 VanessaE there. log out/in.
18:14 Calinou "client disconnected" first time i get this error :P
18:14 VanessaE Calinou: I restarted the server.
18:14 PilzAdam thats not an error
18:15 * Calinou just read some minecon discussion about API: MC devs talk about adding an api in january 2013, then at the end they say "smooth stone stairs are not planned"
18:15 Calinou this is why i prefer minetest... :facepalm:
18:15 Calinou (also, it's not like you can believe them)
18:15 PilzAdam VanessaE, you didnt installed the attachment branch of carts mod
18:15 jin_xi hmm, do you use the attachment branch of carts
18:15 jin_xi ?
18:15 PilzAdam *facepalm
18:16 VanessaE PilzAdam: you didn't say I had to :-)
18:16 Calinou i just noticed z-fighting on the player model
18:16 Calinou (between legs and torso)
18:16 jin_xi afk, back in 10
18:16 Calinou the legs should be moved a tiny bit to prevent that..
18:16 Calinou nothing big though
18:17 VanessaE PilzAdam: what's the branch name>
18:17 PilzAdam attachment
18:17 VanessaE ok, lemme get it.
18:18 * celeron55 has built minetestserver
18:18 VanessaE ok, attachment branch is in.  log out/in.
18:22 VanessaE celeron55: waiting for you to join...
18:22 celeron55 i'm waiting for my server to work
18:22 celeron55 it's broken!
18:22 VanessaE heh
18:24 celeron55 no, it just won't work
18:25 VanessaE well mine's up and four of us are trying to break it, so you suck :D
18:27 celeron55 http://c55.me/random/2012-11/screenshot_953301782.png
18:27 celeron55 ^ mirror
18:28 VanessaE weird!
18:28 PilzAdam i saw this on RBA's server with saplings too
18:30 rsiska joined #minetest-dev
18:30 saschaheylik_ joined #minetest-dev
18:32 VanessaE ok, need a proper "animation" for flying too.
18:32 VanessaE "walking" through the air looks wrong.
18:33 Calinou and too minecrafty 8)
18:33 Calinou swimming maybe?
18:33 Calinou :p
18:33 VanessaE no, "superman" :-)
18:33 VanessaE horizontal, one arm sticking forward, etc. :-)
18:33 PilzAdam there is a better animation mod for MC
18:34 hmmmm oh cool, you guys have player models now
18:34 PilzAdam testing it right now
18:36 VanessaE hmmmm: yeah.  sign on at 24.181.193.82 port 30001 with the models_rebase2.master branch.
18:36 hmmmm can't sign on at the moment
18:36 hmmmm minetest is not in a working state for me
18:44 Taoki celeron55: Back. I will check what's left now
18:45 VanessaE Taoki: test server ^^^
18:45 VanessaE if you wanna play around
18:47 Taoki celeron55: https://github.com/celeron55/minetest_game/commits/models_rebase2.master Looks all good to me
18:47 Taoki So yes, feel free to merge that as well
18:48 PilzAdam Taoki, join the test server
18:48 Taoki will
18:48 VanessaE Taoki: one minor bug, somehow I ended up with a nyan cat rainbow stuck inside me, as an entity.
18:49 PilzAdam also seen with sapling
18:49 Taoki odd, not sure why
18:55 celeron55 i wouldn't call whole model pitching any better than what we currently have
18:55 celeron55 (got the stupid chat sergfault...)
18:55 VanessaE celeron55: how about this:  http://digitalaudioconcepts.com/vanessa/hobbies/minetest/images/default_apple.png
18:55 celeron55 -r
18:56 Taoki I don't think whole model pitching would look too good. That can be done later... it would be nice though
18:56 Taoki But not an emergency so far I believe
18:56 celeron55 VanessaE: that looks unnaturally round, like a christmas tree tingy or a red bomb
18:56 celeron55 thingy*
18:56 VanessaE you mean looking up/down/sideways?
18:57 VanessaE heh
18:57 VanessaE well I suppose I could distort it a bit :-)
18:57 celeron55 punch it at the top
18:58 celeron55 google "red apple"
18:58 VanessaE heh
18:58 VanessaE apples come in all shapes and sizes you know :-)
18:58 hmmmm christmas tree ordimants included
18:59 hmmmm ornament*
18:59 Taoki celeron55: Anyway, as far as I'm concerned both branches are good to go. Do you agree with that, and can you add them upstream now?
19:01 celeron55 i guess i'll push
19:01 PilzAdam btw: https://github.com/celeron55/minetest/commit/0709918bd82ab942445e12aca5bceb85c6ce05e0
19:02 Taoki Awesome. Let me know when it's done, I'm curious to test them from master :D
19:02 Taoki A lot of people will be happy
19:02 VanessaE celeron55: reload the image.
19:02 VanessaE better?
19:02 celeron55 taoki's stuff is now on celeron55/minetest/master and celeron55/minetest_game/master
19:03 Taoki I will update the GIT :)
19:03 celeron55 VanessaE: guess so
19:03 jin_xi yeah yeah yeah
19:03 * VanessaE dances a little gjig
19:03 VanessaE jig*
19:04 Taoki Pulled latest master, compiling now
19:04 PilzAdam celeron55, you read the reminder? https://github.com/celeron55/minetest/commit/0709918bd82ab942445e12aca5bceb85c6ce05e0
19:05 celeron55 uhm, oh that one
19:06 celeron55 doing that now
19:06 VanessaE PilzAdam: pull request for new apple image pending.
19:07 Taoki celeron55: If you're looking at the pull requests, please see also https://github.com/celeron55/minetest/pull/285 It's very small and simple and it was tested, it's working and ready for months
19:07 PilzAdam he already did a rebase branch of that
19:08 Taoki oh ok :)
19:08 Taoki so it's in too?
19:08 PilzAdam https://github.com/celeron55/minetest/tree/anaglyph_rebase1
19:08 Taoki ahh, didn't notice. Will wait then, sure
19:11 PilzAdam ill push my fire branch upstream
19:11 celeron55 pushed
19:12 celeron55 the ContentFeatures thing
19:12 VanessaE will this work against protocol 13?
19:12 VanessaE (e,g. last pre-3d-models git)
19:12 celeron55 not even close 8)
19:13 VanessaE well I meant backward compatible by only showing the 2d stuff.
19:13 VanessaE but okay.
19:13 celeron55 it's not network compatible
19:13 VanessaE too bad.
19:13 celeron55 too much work for too little benefit
19:13 VanessaE ok, this needs a new version number then.
19:14 celeron55 somebody can test and merge the anaglyph stuff if they want
19:14 celeron55 i'm off now ->
19:15 jin_xi nice work
19:15 VanessaE ciao.
19:18 PilzAdam VanessaE, im not sure if the apples are better than the current one
19:18 VanessaE I think they are :-)
19:29 PilzAdam TODO: increase the version number
19:32 Taoki celeron55: later, have fun. Thanks for the help today on the models, it's a wonderful thing they're finally in
19:33 VanessaE ok, server is back and now running upstream.
19:33 Taoki PilzAdam: Wanna look at the anaglyph branch please? It's been tested and working a week ago, but feel free to re-test it too to be sure
19:33 PilzAdam i have no anaglyph glasses, but i will do
19:34 Taoki PilzAdam: Just test that the image is divided between a red and blue channel. I tested with glasses and can confirm that part is well
19:35 PilzAdam first i will join VanessaE's server with upstream client
19:35 Taoki sure thing. Using upstream client too and all is well
19:58 celeron55 oh by the way, here's a good quote: http://naavis.kapsi.fi/gallery/_data/i/upload/2012/11/20/20121120161326-7c3378b2-xx.jpg
19:59 hmmmm many projects have a rather high barrier of entry
20:02 Taoki hehe
20:03 VanessaE heh
20:04 Taoki BTW, new fluid animation directions aren't too good. I'm seeing lava that flows upwards
20:04 PilzAdam yea, sometimes they are wrong
20:07 Taoki BTW, I was curious what you think about another (less important) idea of mine. Can / should health be fully de-hardcoded and put in lua? In builtin perhaps. Since some might want to use their own implementation, while custom mods / games might not want health at all but maybe a different mechanic
20:07 Taoki And IMO engine shouldn't handle any rules that could be game specific
20:07 Taoki A builtin script could hold a health value, and remove an entity accordingly if it reaches 0
20:11 hmmmm Taoki, yes
20:11 hmmmm I mentioned this idea a while back, scan the logs for "armor bar" or something on that order
20:12 hmmmm i mentioned what had to be done and what not
20:12 Taoki nice
20:12 Taoki we need lua hud items first, but i mentioned this and it's a feature that should come eventually too
20:13 hmmmm http://irc.minetest.ru/minetest-dev/2012-11-05
20:15 hmmmm this is really easy to implement though
20:15 hmmmm just need to make those three new packets, the handlers, the lua bindings,
20:28 Taoki joined #minetest-dev
20:31 celeron55 hmmmm: how is the mapgen stuff going?
20:32 celeron55 i am a bit worried that you will quit without releasing anything
20:32 hmmmm no i will not quit
20:32 hmmmm it's fine, it's just that i went to a new computer and i have to resetup everything so that's taking a bit of time
20:32 hmmmm right now it compiles but doesn't generate anything, i need to debug it
20:33 celeron55 it could be involuntary too, you might eg. die, and nobody could benefit at all of anything you've done
20:33 hmmmm isn't that a bit of a faux pas, you know, committing something to a repository that doesn't work?
20:34 celeron55 everybody else than you do that :P
20:34 celeron55 including me
20:34 hmmmm alright i'll commit it
20:35 celeron55 if you're worried of people trying to use it as if it'd work, make it a non-master branch so they can suspect it's something experimental
20:35 hmmmm it's in my experience that if something isn't 100% when it's released to the public, people will try to use it and people will complain that it doesn't work
20:35 hmmmm and i'd get 1000 messages "hey your thing crashed when i did blah blah blah"
20:35 hmmmm like, yes i know that!
20:36 celeron55 putting stuff in branches on github doesn't mean any of the general public will see it
20:36 celeron55 i currently (like always) have a bunch of random experiemntal branches on github and haven't seen anyone complain about them
20:36 celeron55 and that is as public as a branch in a minetest repo can get
20:37 celeron55 (unless you advertise one somewhere :P)
20:37 hmmmm hrmm
20:38 celeron55 think of it more like a public backup on the internet
20:38 Taoki What mapgen stuff are you working on?
20:38 hmmmm how do i see a preview of my commit
20:38 hmmmm taoki, a completely new framework
20:38 Taoki Awesome
20:38 hmmmm also in the plans is to fix up the old mapgen and have that as an option
20:38 Taoki I assume I'm not that lucky to see lua biomes as well with it :P
20:39 hmmmm i'd just need to make Mapgen an interface
20:39 Taoki and removal of hard-coded ones
20:39 hmmmm oh of course, the biome thing was sort of the central point
20:39 Taoki oh? So now biomes are defined in lua script?
20:39 hmmmm yes
20:39 Taoki Beautiful! Been waiiting a while for that :D
20:39 Taoki Wasn't hoping of seeing it happen soon
20:40 hmmmm this was supposed to be released on halloween
20:40 hmmmm guess how well that worked out
20:40 Taoki celeron55: BTW. What is your take on using LevelDB instead of sqlite? OldCoder said it's both smaller and faster, and if so it might be nice to put it in upstream too
20:40 Taoki hmmmm: So when will it be on GIT?
20:41 hmmmm i just need to go through and see exactly what's going to be committed
20:41 hmmmm and then i'll hit the button
20:41 celeron55 Taoki: i haven't looked much into that
20:41 hmmmm for some reason it seems like it's removing all the current lines in all the files and readding them
20:42 Taoki ok
20:42 celeron55 hmmmm: if you're not too good using git, make sure to have backups 8)
20:42 hmmmm could be a newline thing?
20:42 hmmmm i tried testing this under MSVC
20:42 celeron55 git will gladly remove all of your work if you accidentally ask it to
20:42 hmmmm I have 3 backups of this
20:43 Taoki BTW... something a littlle odd. Since I moved to Linux I occasionally experience plasma desktop crashing, either freezing my machine or loggiing me out. I get the slight impression this only happens after I run minetest. I wonder if it might have some flag in the video driver that messes the proprietary ATI driver up. Very hard to debug those currently
20:43 celeron55 it takes a certain amount of experience to get to the level of not needing backups with git
20:43 Taoki hmmmm: Before adding to git, i recommend copying your src folder to a backup location to be safe
20:44 celeron55 hmmmm: just go on and try what happens, and if it goes wrong, see what happened and try again somehow differently 8)
20:45 hmmmm when i do git diff, it compares with the stuff in HEAD, FETCH_HEAD, or ORIG_HEAD?
20:48 celeron55 umm... what do those even mean 8)
20:49 celeron55 i don't think ORIG_HEAD and FETCH_HEAD are relevant at all
20:50 Taoki I think HEAD is the latest revision on the local branch you're currently on, FETCH_HEAD the latest rev on the branch you're fetching, and ORIG_HEAD might be the origin you're pulling from
20:50 celeron55 if you run "git diff", it will get the diff from HEAD to the current contents of files
20:50 hmmmm i want to do a binary comparison against HEAD
20:50 hmmmm where are the actual files from HEAD stored in .git?
20:50 celeron55 if you run "git diff <commit>" you will get the diff from the commit to the current content of files... and with two commits it's obvious
20:50 celeron55 git doesn't store files
20:50 celeron55 it stores some more abstract thingies
20:51 hmmmm i am pretty sure this is a newline problem thanks to msvc
20:51 celeron55 use git status, git log, and even better, gitk or something to inspect what you have
20:52 celeron55 or just push it to github and let others see :P
20:52 hmmmm ergok
20:52 hmmmm okay
20:52 Taoki Yes, push to github is best
20:53 hmmmm whoops
20:53 hmmmm neeed to configure name and email
20:54 Taoki yeah, GIT wants that first. You'll also need to provide your github user and password when pushing
20:55 hmmmm which i forgot
20:55 hmmmm do you recall if github requires "at least 9 characters for password" or something on that order?
20:56 Taoki not really. Might require at least 6 though
20:56 Taoki Or 8 or something, dunno
20:56 celeron55 i don't think github even allows pushing with a password
20:56 celeron55 it always wants a public ssh key
20:57 hmmmm agh i'll just reset my password
20:57 celeron55 read the documentation at there
20:57 Taoki for me it uses username and password
20:58 celeron55 ...well, then i guess it allows that too :P
20:58 Taoki hmmmm: Please backup the src folder with your changes... just copy it somewhere. It's safer that way just to be sure git can't mess something up
20:58 celeron55 he already said he has backups
20:58 hmmmm yep
20:59 Taoki ah, sorry
21:00 hmmmm ahh i finally remembered my gmail password
21:01 hmmmm i ought to change my scheme to SHA1(website:username:some salt)
21:03 hmmmm and of course only biome.cpp got uploaded
21:07 hmmmm so i think i've got an idea
21:08 hmmmm remove that last commit, nuke everything locally, git clone, fix all newlines in the current source files, copy them over to the newly cloned repository, THEN merge
21:22 VanessaE I think themipmap/aniso code got broken in latest git, doesn't seem to work for me anymore :(
21:24 VanessaE had to restart the client to make it work - that's a regression.
21:26 Taoki celeron55: BTW, in case this isn't known yet. In multiplayer, you can hear other player's footsteps, but not when they dig or place nodes. Which is odd knowing the LUA architecture... sounds are sent to every client
21:28 VanessaE Taoki: to avoid linear filtering artifacts on the player models, you should pad each texture by one pixel, just copy whatever the pixels are at each edge.
21:28 Taoki Dunno how to do that, and that sounds like it might mess with the texture even more. Sadly I can't help there
21:29 Taoki But it's not a very huge issue I guess. It happens on nodes too
21:29 VanessaE yeah I know
21:30 VanessaE the other solution is high-resolution textures for players.  I take it the existing minecraft format can simply be upsized?
21:30 jin_xi graphics setting change works here with exit to menu alone
21:31 Taoki VanessaE: Sure. RBA altrady tested that
21:31 VanessaE jin_xi: ok, must have been a glitch on my end.
21:31 VanessaE Taoki: ok, cool.
21:31 Taoki Obviously not for default minetest game, but custom games can have highres textures
21:31 Taoki Important thing to note is it can have high-poly models too. You CAN put animated players of the quality of very modern games in
21:33 Taoki hmmmm: How's it going with that GIT? did you succeed?
21:41 hmmmm i'm not constantly working on stuff, i just got back from dealing with RL things
21:43 Taoki oh, sure then
21:44 hmmmm oh my god i can't stand it
21:44 hmmmm why do people whine and refuse to help themselves
21:47 celeron55 because most people are dumb
21:47 celeron55 i wonder how one can't already be used to that :D
21:47 celeron55 and let me rephrase that: most people are *incredibly* dumb
22:06 rsiska joined #minetest-dev
22:17 rsiska joined #minetest-dev
22:39 sstrandberg joined #minetest-dev
22:43 celeron55 i think i need to release 0.4.4 asap
22:44 VanessaE yes.
22:44 VanessaE or at the very least, bump the version number on what we have now, it differs enough that people, stupid as they are, will be confused :-)
22:49 celeron55 hmm
22:49 * Taoki wonders how far hmmmm got with that lua mapgen, since I'm very interested in the feature
22:50 celeron55 he's been working on it for, like, months
22:51 Taoki awesome
22:51 celeron55 i think i'll bump the version, because the current one isn't compatible with 0.4.3, and is likely to be compatible with 0.4.4
22:51 Taoki I'm waiting to see it on GIT, but I'm not sure if he's back online and working on MT again
22:54 celeron55 hmmmm doesn't seem to tend to rush too much 8) i guess we'll see some WIP stuff in git some day
22:58 hmmmm I'm working on it
22:58 hmmmm having lots of trouble with this  http://ideone.com/bLhBuv  any ideas?
22:58 hmmmm it's supposed to strip all \r out of source files, this is my main problem (i stupidly opened it up in visual studio and it forcibly changed all my line endings)
22:59 hmmmm that's why the entire file doesn't match with diff
22:59 VanessaE can't just pipe it through tr?
22:59 VanessaE or sed?
22:59 hmmmm nope hahahaha
22:59 VanessaE damn :-)
22:59 hmmmm it'll just erase the whole file
22:59 hmmmm has something to do with buffering
23:00 VanessaE hah, didn't see you're already using it.
23:00 VanessaE or trying to.
23:01 hmmmm i've wasted like 30 minutes trying to get this to work
23:01 hmmmm ridiculous
23:01 hmmmm such a simple task
23:01 VanessaE why are you overwriting the original file?
23:01 VanessaE I don't think tr does that by default.
23:01 hmmmm why wouldn't i?
23:01 hmmmm alright, i'll try something different
23:02 VanessaE cat $i |tr -d \\015 >$i.tmp; mv $i.tmp $i
23:02 VanessaE maybe something like that is safer.
23:02 celeron55 hmmmm: if you are working on windows, be aware that git also has some very stupid automatic newline conversion functionality which you should always have switched off for best results
23:03 celeron55 (it defaults to being on some crazy "auto" mode)
23:04 hmmmm it's the visual studio IDE itself that screwed this up
23:04 hmmmm i just open it once
23:04 celeron55 (actually, that also applies for other OSes)
23:04 hmmmm and all the line endings are changed
23:04 hmmmm it would actually be nice if git justified them to \n before doing the diff
23:05 celeron55 http://wiki.opf-labs.org/display/SP/Configuring+how+line-endings+are+handled+by+git
23:05 celeron55 it has those options
23:06 celeron55 also here is some random stuff http://stackoverflow.com/questions/2517190/how-do-i-force-git-to-use-lf-instead-of-crlf-under-windows
23:06 celeron55 you can screw it up or fix it (by choice or by luck) at pull or push time
23:07 hmmmm a.. ha
23:08 hmmmm on my system, core.autocrlf isn't set, which is why it tried to upload with CRLF
23:08 hmmmm so i actually have to enable this setting and then hopefully it'll work
23:08 celeron55 that might work
23:09 celeron55 the automatic setting doesn't really follow the "least surprise" way though; everybody definitely would need to know that that stuff exists
23:10 celeron55 for example, if you have newline conversions set on a windows machine pushing stuff, you may end up pushing newlines as \n to a .bat file that you have modified
23:11 celeron55 then if you pull that with a linux machine (or a windows one with conversions turned off), you get an invalid .bat file (that matters on linux when you want to eg. cross-compile and/or package for windows)
23:11 celeron55 (.bat doesn't work with unix newlines 8))
23:13 celeron55 the only way that always works is to have all conversions turned off and by setting up your editor to use the appropriate newlines at the appropriate times
23:14 celeron55 (been there, done that)
23:19 rob_ joined #minetest-dev
23:23 Taoki hmmmm: Did you get the Github username & password or ssh key working? If it's user and pass i might be able to help since I use that, though generaly I'm not that great at GIT
23:45 ssstrandberg joined #minetest-dev
23:54 hmmmm eeyy
23:54 hmmmm just got finished eating, the last thing i was going to try works great
23:55 VanessaE well dessert tends to be like that.
23:55 VanessaE next time try adding some of those nutty crunchy sprinkles.
23:55 hmmmm I am not messing around the the autocrlf setting in git
23:55 hmmmm i'd rather convert them manually and then not mess with anything else
23:56 hmmmm alright, well unfortunately it seems that absolutely _all_ of the source files were converted to crlf
23:56 hmmmm so i should probably be using find instead of foreach *.cpp
23:56 Taoki Does that break anything? At least you have backups if so
23:56 * Taoki never really understood the whole clrf thing
23:56 hmmmm dude i have backups galore
23:57 hmmmm too many backups
23:57 Taoki yeah
23:57 Taoki :P
23:58 Taoki What happens if the files get converted to clrf? Does it break something?
23:58 Taoki Like convert tabs to spaces
23:59 SpeedProg joined #minetest-dev
23:59 hmmmm well here's what happened to me exactly:  i went to go test my work on windows, so i opened up the project in MS visual studio, visual studio converts all your line endings from LF to CRLF

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