| Time |
Nick |
Message |
| 00:02 |
Megaf |
RealBadAngel, meh, thats indeed a block of useless code |
| 00:03 |
Megaf |
kahrl, suppose we remove what RealBadAngel suggested and client tried to connect to old server, worst case scenario, what would happen? |
| 00:04 |
RealBadAngel |
kahrl, http://i.imgur.com/1IvaaF3.png |
| 00:04 |
RealBadAngel |
^^ i can see wielded glitch on a system that was previously non affected |
| 00:05 |
RealBadAngel |
only difference now is that im rendering to texture instead of directly |
| 00:05 |
kahrl |
Megaf: I think you'd get lots of "WARNING: ClientActiveObject: No factory for type=..." and won't be able to see or interact with dropped items |
| 00:05 |
Fixer |
RealBadAngel, yes, that is it |
| 00:06 |
Megaf |
RealBadAngel, something similar happens to me when I hold a boat |
| 00:06 |
Megaf |
but the axe looks normal |
| 00:06 |
Fixer |
RealBadAngel, please find nearest snow biome with forest |
| 00:06 |
Megaf |
(shaders off) |
| 00:06 |
kahrl |
RealBadAngel: interesting |
| 00:06 |
RealBadAngel |
ok |
| 00:06 |
kahrl |
what is the difference between rendering to screen and rendering to texture? |
| 00:06 |
Fixer |
I've tried it with literally ALL settings, and bug won't go away |
| 00:06 |
Fixer |
including changing driver settings |
| 00:07 |
Megaf |
and for me the thing on http://i.imgur.com/1IvaaF3.png is not even a bug, just a bummaping thing... |
| 00:07 |
Fixer |
RealBadAngel, find some tree with snow on it, go above it, increase height, move mouse a lit, look when ripples on snow appear |
| 00:07 |
RealBadAngel |
Fixer, ok |
| 00:08 |
RealBadAngel |
ok i can see it |
| 00:08 |
RealBadAngel |
hold on a sec |
| 00:08 |
kahrl |
Megaf: that would be a pretty crappy bump map if you ask me :P |
| 00:08 |
Fixer |
RealBadAngel, finally |
| 00:09 |
RealBadAngel |
http://i.imgur.com/ozXnPCT.png |
| 00:09 |
Fixer |
RealBadAngel, can you post a pic? |
| 00:09 |
RealBadAngel |
kahrl, those are not shaders related things at all |
| 00:09 |
Fixer |
RealBadAngel, YES |
| 00:09 |
Fixer |
RealBadAngel, that is that s--t |
| 00:09 |
Fixer |
confirmed |
| 00:09 |
sofar |
ugh crap, backface culling doesn't work for mesh nodes |
| 00:09 |
Fixer |
RealBadAngel, so weild mesh problem is also related to texture tear, it seems |
| 00:10 |
Fixer |
or it is the same problem actually |
| 00:10 |
Fixer |
RealBadAngel, if you go higher above trees they will look more horrible |
| 00:10 |
Fixer |
crappiness increases with distance |
| 00:10 |
RealBadAngel |
i believe we set the materials somehow wrong way |
| 00:10 |
RealBadAngel |
since now i can see the glitches i will find a reason |
| 00:11 |
Megaf |
kahrl, low resolution, I never fully undestood all shaders and mapping for minetest |
| 00:11 |
Megaf |
with such low res textures |
| 00:11 |
Megaf |
minetest was never abut eye candy |
| 00:11 |
RealBadAngel |
Megaf, actual texture resolution is not important in opengl |
| 00:12 |
RealBadAngel |
texture coords are always 0-1, no matter its size |
| 00:12 |
RealBadAngel |
so i can shade 1px texture or 2048x the very same way |
| 00:12 |
Megaf |
now, speaking of OpenGL, couldnt we drop OpenGL and Direct3Dx and use only OpenGL ES 2.x? |
| 00:12 |
RealBadAngel |
no we cant |
| 00:13 |
Megaf |
about texture size, the memory use changes |
| 00:13 |
Megaf |
GPU memory use |
| 00:13 |
RealBadAngel |
ES is too limited |
| 00:13 |
Megaf |
are you sure we need stuff from OpenGL? |
| 00:13 |
RealBadAngel |
ES is meant for mobile stuff, not regular PC and powerful GPUs |
| 00:13 |
Megaf |
I have seen lots of ES games and they look pretty awesome |
| 00:14 |
Megaf |
and if it is menat for mobile, I think its more efficient |
| 00:14 |
RealBadAngel |
ive developed ES version of shaders |
| 00:14 |
sapier |
megaf it's hard enough to get working opengl drivers on some windows machines ;-) I doubt it'll be better if we'd use opengles |
| 00:14 |
RealBadAngel |
we can have both versions |
| 00:15 |
Megaf |
sapier, maybe if we used only a single version of OpenGL then? like 3.1 or 2.x |
| 00:15 |
RealBadAngel |
sofar, you need backface culling for meshes? |
| 00:15 |
Megaf |
(PS, I don't know what I'm talking about in case you didn't notice) |
| 00:15 |
sapier |
for what I understand most of that stuff is done by irrlicht (except shaders of course) |
| 00:16 |
sapier |
and as of windows there's still some gpu's out there where you can only reasonably use directx ... and I guess it's gonna get even worse as pc market started to fragment even more since microsoft force installs win10 |
| 00:16 |
sofar |
RealBadAngel: it's hardcoded disabled in minetest/nodedef.cpp#912 |
| 00:17 |
RealBadAngel |
https://github.com/minetest/minetest/pull/3391 |
| 00:17 |
RealBadAngel |
sofar, this PR allows to define backface culling for each tile |
| 00:17 |
sofar |
I don't need it per tile |
| 00:17 |
sofar |
of course that's not a bad idea |
| 00:17 |
RealBadAngel |
mesh textures are stored as a tile |
| 00:18 |
sofar |
but I have only 1 texture |
| 00:18 |
RealBadAngel |
so 1 tile |
| 00:18 |
sofar |
IC |
| 00:18 |
RealBadAngel |
then you could set the flags for that tile (material) |
| 00:18 |
sofar |
all I want is backface_culling = true to apply to mesh nodes |
| 00:18 |
RealBadAngel |
heres the code that does that and a few more things |
| 00:19 |
RealBadAngel |
happily was rejected because of ability to force filters |
| 00:19 |
Megaf |
RealBadAngel, I removed the code you said about CAO, could connect to my server, and my server is a now old build |
| 00:19 |
Megaf |
no problem with dropped items |
| 00:20 |
RealBadAngel |
Megaf, you should propably try to connect some ancient build |
| 00:20 |
RealBadAngel |
something like at least 1-1,5 yrs old |
| 00:20 |
Megaf |
meh, who uses that? |
| 00:21 |
Megaf |
I actually have a 3 or 4 years old backup of an old server |
| 00:21 |
Megaf |
somewhre |
| 00:21 |
Megaf |
anyway, I believe we shouldnt worry about that |
| 00:27 |
|
Lunatrius joined #minetest-dev |
| 00:38 |
sofar |
RealBadAngel: there's so many arbitrary restrictions on node visual attributions :( |
| 00:39 |
RealBadAngel |
i havent invented that |
| 00:39 |
sofar |
can't we make mesh nodes default to backface_culling = false |
| 00:39 |
sofar |
and allow nodedef's to enable it? |
| 00:39 |
sofar |
I tried but I can't seem to get that to work |
| 00:39 |
RealBadAngel |
best way is to enable defining that by mods |
| 00:39 |
sofar |
that's what I want to |
| 00:39 |
sofar |
but it's hardcoded disabled for mesh nodes |
| 00:39 |
sofar |
which is dumb |
| 00:40 |
sofar |
I mean, if it looks like shit, people won't use it |
| 00:40 |
sofar |
but in my new door case, I need it, and it works |
| 00:40 |
sofar |
it even shows a problem with one of my meshes |
| 00:41 |
RealBadAngel |
either we can set default to false (but then propably some1 would like it to true) |
| 00:41 |
RealBadAngel |
or merge my patch |
| 00:41 |
sofar |
yeah but, this should be a 2-line change |
| 00:41 |
RealBadAngel |
then the problem is gone |
| 00:41 |
sofar |
your patch is... not a 2-liner |
| 00:41 |
sofar |
2-liners are easy to merge |
| 00:41 |
RealBadAngel |
what if you will need backface culling set to true? |
| 00:42 |
RealBadAngel |
you will make next 2 liner or revert previous one? ;) |
| 00:42 |
sofar |
https://gist.github.com/sofar/971b26ffbff16e1183d4 |
| 00:42 |
sofar |
I was thinking something like that |
| 00:42 |
sofar |
default to off, enable when mod tells it to |
| 00:43 |
RealBadAngel |
but theres no way to do that without my patch |
| 00:43 |
sofar |
I see backface culling work |
| 00:43 |
sofar |
with my patch |
| 00:43 |
sofar |
I just can't turn it OFF |
| 00:44 |
sofar |
oh, hmm |
| 00:44 |
RealBadAngel |
what? |
| 00:49 |
sofar |
So ContentFeatures enables backface_culling at object creation |
| 00:49 |
sofar |
nodedef.cpp disables it when NDT_MESH |
| 00:50 |
sofar |
https://gist.github.com/sofar/363ea387e054f88c2b0c |
| 00:50 |
sofar |
I'm thinking, once we see the "drawtype" == NDT_MESH, we set it to false |
| 00:50 |
sofar |
and if it's then read to be true, enable it |
| 00:51 |
sofar |
oh hum, probably needs to be serialized/unserialized too |
| 00:51 |
RealBadAngel |
theres a case that already let that |
| 00:51 |
RealBadAngel |
see water node defs |
| 00:52 |
RealBadAngel |
my patch just extends that for regular tiles |
| 00:53 |
sofar |
got it to work |
| 00:53 |
sofar |
yeah I understand your patch |
| 00:54 |
|
sapier1 joined #minetest-dev |
| 00:57 |
RealBadAngel |
funny, that texture tear is visible on top of snow slab, but not on regular snow node |
| 01:00 |
sofar |
ugh why does blender make such a mess of meshes when I export them? |
| 01:00 |
sofar |
somehow even recalculating normals it just doesn't do the right thing |
| 01:00 |
|
DFeniks joined #minetest-dev |
| 01:03 |
RealBadAngel |
blender is weird, i was fighting with it a few hours lately to rescale and rotate stargate mesh |
| 01:03 |
RealBadAngel |
this is really a tool for geeks |
| 01:04 |
kaadmy |
if you know it ;) |
| 01:13 |
|
luizrpgluiz joined #minetest-dev |
| 01:20 |
luizrpgluiz |
which would be responsible for leaving the file faster minetest? |
| 01:24 |
sofar |
luizrpgluiz: I can't make sense of that sentence |
| 01:26 |
luizrpgluiz |
Paramat said that in minetest, when you leave the game options without leaving the most beautiful game like shaders, he has loss 20-70% drop frames per second |
| 01:27 |
sofar |
right |
| 01:28 |
luizrpgluiz |
I had an idea to try to resolve this issue it would load the shader as a non-binding option |
| 01:30 |
luizrpgluiz |
because disabling all the graphics options of the game, it was to have a very good improvement in frames per second |
| 01:33 |
sofar |
next up: replace chests with an animating chest! |
| 01:33 |
* sofar |
giggles |
| 01:44 |
|
luizrpgluiz left #minetest-dev |
| 01:53 |
RealBadAngel |
70% fps drop, wow |
| 01:54 |
RealBadAngel |
paramat is doing here excellent job at making chaos |
| 01:55 |
sofar |
he's being misquoted |
| 01:55 |
sofar |
please |
| 01:56 |
RealBadAngel |
whatever |
| 01:56 |
RealBadAngel |
fiddling now with wielded glitch, cant find a difference why the same node is displayed properly as entity and wrong in hand |
| 01:58 |
RealBadAngel |
im not sure whats that in fact, frontface culling? |
| 01:58 |
sofar |
hand culling |
| 02:00 |
RealBadAngel |
but noticed something else |
| 02:00 |
RealBadAngel |
http://i.imgur.com/izVO2Ex.png |
| 02:01 |
sofar |
why is it fuzzy? |
| 02:01 |
RealBadAngel |
it looks like wielded mesh is somehow broken anyway, see that cube attached to the original one? |
| 02:01 |
RealBadAngel |
fuzzy, because client started in window and then i rescaled it |
| 02:02 |
RealBadAngel |
posprocessing works on the resolution set at the start, havent fixed that yet |
| 02:03 |
RealBadAngel |
also im trying all the possible material flags one by one |
| 02:03 |
|
behalebabo left #minetest-dev |
| 02:04 |
RealBadAngel |
i think i will try another wield mesh generator |
| 02:06 |
RealBadAngel |
heh |
| 02:07 |
RealBadAngel |
looks like also cubes are displayed wrong way |
| 02:07 |
RealBadAngel |
ive made series of screenshots to catch swinging the wield item |
| 02:08 |
RealBadAngel |
http://i.imgur.com/c62Jt8y.png |
| 02:09 |
|
kaadmy joined #minetest-dev |
| 02:09 |
Fritigern |
Looks like the faces are displayed inside-out |
| 02:10 |
RealBadAngel |
or some faces are missing |
| 02:11 |
sofar |
did you mess up the normals? ;) |
| 02:11 |
RealBadAngel |
no |
| 02:11 |
RealBadAngel |
maybe i should recalculate them |
| 02:12 |
|
kaadmy joined #minetest-dev |
| 02:24 |
RealBadAngel |
that doesnt help |
| 02:39 |
|
paramat joined #minetest-dev |
| 02:43 |
paramat |
hmmmm (or anyone) minetest.place_schematic_on_vmanip() seems to cause a memory leak https://github.com/minetest/minetest/issues/3566 |
| 02:57 |
|
sapier1 left #minetest-dev |
| 03:11 |
hmmmm |
how was it determined to be the API itself? |
| 03:11 |
hmmmm |
what if it's an error in the mod, holding onto references of the vmanip? |
| 03:12 |
paramat |
it could be the mod |
| 03:12 |
paramat |
i couldn't see an error though |
| 03:13 |
hmmmm |
could you see an error in l_place_schematic_on_vmanip? |
| 03:13 |
paramat |
not yet |
| 03:15 |
hmmmm |
the standard method of determining a memory leak is by running minetest under valgrind |
| 03:15 |
hmmmm |
i'd start with that |
| 03:15 |
paramat |
ah ok |
| 03:58 |
|
MyTeke_ChrisWMas joined #minetest-dev |
| 04:12 |
sofar |
heh, how well does it run under valgrind? I should try... |
| 04:25 |
hmmmm |
in any case, i'm skeptical of the assertion that the API is bugged because it doesn't make any allocations except for schematic loading, but that's managed by the schematic manager and has very well-defined lifetimes |
| 04:26 |
hmmmm |
it's much more likely that the mod is bugged by holding onto references or the author doesn't realize the GC just hasn't run yet |
| 04:29 |
sofar |
holy crap can we tell valgrind to ignore luajit? |
| 04:29 |
hmmmm |
yes, there is a way to do this |
| 04:30 |
sofar |
yeah, I know ... just haven't done it myself |
| 04:32 |
|
paramat joined #minetest-dev |
| 04:33 |
sofar |
well I'm glad my culling change is getting a quick thumbs up already |
| 04:33 |
sofar |
otherwise I would have seen the doors rewrite strand pretty quickly |
| 04:41 |
sofar |
possibly lost:124 bytes |
| 04:41 |
sofar |
well now, that doesn't look so bad |
| 05:00 |
sofar |
https://github.com/dear-github/dear-github |
| 05:38 |
|
lezzy joined #minetest-dev |
| 05:52 |
|
Player2 joined #minetest-dev |
| 06:08 |
|
lezzy joined #minetest-dev |
| 06:09 |
paramat |
'Liquid flow: Prevent water spreading on ignore WIP' https://github.com/minetest/minetest/pull/3581 |
| 06:10 |
sofar |
that's some progress |
| 06:19 |
paramat |
hopefully. the abm i tried worked but to be lightweight would need a low chance, and therefore would take a while to have effect |
| 06:22 |
sofar |
it doesn't need to fire often |
| 06:22 |
sofar |
who cares if water takes a bit to self-repair? |
| 06:22 |
paramat |
exactly |
| 06:23 |
paramat |
i'll look into a new API for adding a liquid node to the liquid queue |
| 07:21 |
|
PenguinDad joined #minetest-dev |
| 07:29 |
|
blaze joined #minetest-dev |
| 07:48 |
|
nrzkt joined #minetest-dev |
| 07:52 |
paramat |
ahh there is already an API for adding a node to the liquid queue =) |
| 08:04 |
paramat |
doesn't work, back to plan A |
| 08:07 |
paramat |
actually it does, nice |
| 08:10 |
|
Hunterz joined #minetest-dev |
| 08:14 |
|
Krock joined #minetest-dev |
| 08:39 |
|
paramat left #minetest-dev |
| 09:16 |
|
Obani joined #minetest-dev |
| 09:16 |
|
blaze joined #minetest-dev |
| 09:53 |
|
Calinou joined #minetest-dev |
| 09:57 |
|
CraigyDavi joined #minetest-dev |
| 10:34 |
|
CraigyDavi joined #minetest-dev |
| 10:54 |
|
Megaf joined #minetest-dev |
| 11:43 |
|
Fixer joined #minetest-dev |
| 11:57 |
|
Obani joined #minetest-dev |
| 12:00 |
|
H-H-H joined #minetest-dev |
| 12:53 |
|
rubenwardy joined #minetest-dev |
| 13:11 |
|
Amaz joined #minetest-dev |
| 13:31 |
|
jin_xi joined #minetest-dev |
| 13:55 |
|
Obani joined #minetest-dev |
| 14:41 |
|
nrzkt joined #minetest-dev |
| 14:55 |
|
turtleman joined #minetest-dev |
| 15:10 |
|
hmmmm joined #minetest-dev |
| 15:11 |
|
loggingbot[ joined #minetest-dev |
| 15:25 |
|
Calinou joined #minetest-dev |
| 15:31 |
|
blaze joined #minetest-dev |
| 15:32 |
|
kaadmy joined #minetest-dev |
| 17:02 |
|
kaeza joined #minetest-dev |
| 17:07 |
|
Alduin_ joined #minetest-dev |
| 17:07 |
|
nolsen joined #minetest-dev |
| 17:09 |
|
nolsen joined #minetest-dev |
| 17:27 |
|
mangeurdenuage joined #minetest-dev |
| 17:29 |
|
H-H-H joined #minetest-dev |
| 17:32 |
|
Obani joined #minetest-dev |
| 17:51 |
|
H-H-H joined #minetest-dev |
| 18:45 |
|
loggingbot_2 joined #minetest-dev |
| 18:45 |
|
Topic for #minetest-dev is now Minetest core development and maintenance. Last release: 0.4.13, Aug 20 2015. Chit-chat goes to #minetest. Consider this instead of /msg celeron55. http://irc.minetest.ru/minetest-dev/ http://dev.minetest.net/ |
| 18:46 |
|
Elinvention joined #minetest-dev |
| 18:58 |
|
PenguinDad joined #minetest-dev |
| 19:39 |
|
everamzah joined #minetest-dev |
| 19:58 |
|
paramat joined #minetest-dev |
| 19:59 |
|
Megaf joined #minetest-dev |
| 20:05 |
|
Obani joined #minetest-dev |
| 20:09 |
paramat |
can anyone review this simple improvement for water on ignore? https://github.com/minetest/minetest/pull/3581 |
| 20:10 |
|
Elinvention joined #minetest-dev |
| 20:10 |
paramat |
will merge 3578 later |
| 20:37 |
Fixer |
paramat, interesting, is ABM for liquid activation included in that patch? |
| 20:38 |
paramat |
no, the ABM is hacky and experimental, and would be in mtgame anyway |
| 20:39 |
paramat |
the ABM could be added by anyone in a mod though |
| 20:40 |
Fixer |
nah, please make it included, those horrible liquid bugs should be fixed away from MT entirely |
| 20:43 |
paramat |
if i can make the ABM work well i'll open a PR for MTGame |
| 20:46 |
Fixer |
do you need some world 'production' copy with liquid bugs for testing? |
| 20:47 |
paramat |
no thanks, luckily liquid bugs are super easy to create =) |
| 20:48 |
Fixer |
try adding several water source and dump them over each other in different places ._. |
| 20:48 |
Fixer |
i guess you know it anyway |
| 20:48 |
Megaf |
I wonder if I could get the source code for all libs that the buildbot download and build them myself |
| 20:49 |
Fixer |
paramat, there was also liquid bug that thin diagonal water 'slab' remained after some exting-ed flowes |
| 20:49 |
Fixer |
i have no idea wth was that |
| 20:51 |
Fixer |
paramat, ABM could be also tested in valleys caves that got planty of stuck lava over ignore, seen it yesterday a lot |
| 20:51 |
paramat |
ah yes |
| 20:51 |
|
leat joined #minetest-dev |
| 20:52 |
Fixer |
with fixed flows that caves will be so hardcore to explore |
| 20:58 |
sofar |
paramat: stuck with kiddos for a bit, can maybe test later |
| 20:58 |
paramat |
no prob |
| 21:03 |
|
iamafriend5 joined #minetest-dev |
| 21:28 |
|
sfan5_ joined #minetest-dev |
| 21:29 |
|
paramat left #minetest-dev |
| 21:59 |
|
RealBadAngel joined #minetest-dev |
| 22:00 |
RealBadAngel |
kahrl, here? |
| 22:04 |
RealBadAngel |
wielded item glitch basically is zbuffer not working i think |
| 22:05 |
RealBadAngel |
changing order of planes in extruded mesh buffers changes the visual effect |
| 22:05 |
RealBadAngel |
if we add z+ and z- planes at the end of the buffer instead of at the start we get another looking wiedled item |
| 22:07 |
RealBadAngel |
http://i.imgur.com/8fiu26Z.png |
| 22:08 |
RealBadAngel |
side of the item is now lookin correct |
| 22:08 |
RealBadAngel |
but notice the waves on the inner one |
| 22:09 |
RealBadAngel |
http://i.imgur.com/DxmgIby.png |
| 22:09 |
RealBadAngel |
especially visible on hand |
| 22:10 |
|
Sokomine joined #minetest-dev |
| 22:20 |
|
rubenwardy joined #minetest-dev |
| 22:47 |
|
werwerwer joined #minetest-dev |
| 22:50 |
|
johnnyjoy joined #minetest-dev |
| 22:55 |
|
Ritchie joined #minetest-dev |
| 22:57 |
|
rubenwardy joined #minetest-dev |
| 22:59 |
|
nolsen joined #minetest-dev |
| 23:03 |
|
est31 joined #minetest-dev |
| 23:11 |
|
Fixer joined #minetest-dev |
| 23:14 |
|
mangeurdenuage left #minetest-dev |
| 23:33 |
|
DFeniks joined #minetest-dev |
| 23:44 |
|
nolsen joined #minetest-dev |
| 23:54 |
est31 |
anybody want to fix https://github.com/minetest/minetest/issues/3512 |
| 23:54 |
est31 |
build failure is big bug, supporting future compilers is as important as supporting old ones. |
| 23:54 |
sofar |
c++11 isn't something we want to avoid either |
| 23:55 |
est31 |
keeping the bug in master ruins bisect once we switch to c++11 |
| 23:55 |
est31 |
requiring c++11 != being able to build with -std=c++11 |
| 23:55 |
RealBadAngel |
btw, i think i know the real reason for wielded glitch |
| 23:56 |
RealBadAngel |
im pretty sure i will have working fix soon |
| 23:56 |
RealBadAngel |
extruded mesh was broken |
| 23:57 |
est31 |
if anybody wants to reproduce the build failure, run cmake with -DCMAKE_CXX_FLAGS="-std=c++11" |
| 23:59 |
est31 |
I'll keep the offer to fix the bug around for one week, after that I'll try a fix myself |
