| Time |
Nick |
Message |
| 00:00 |
gregorycu |
Well written python code looks very simple |
| 00:00 |
gregorycu |
But under the covers it is not |
| 00:01 |
cmdskp |
Just don't get me started on COBOL, Fortran or the hideous Prolog language...*shudders* |
| 00:01 |
gregorycu |
I've decided to branch minetest and add python scripting support, just haven't got around to it :) (Not expecting this work to be merged into main) |
| 00:01 |
cmdskp |
that last one wraps your head up and twists your spinal column inside-out and recurses it backwards... |
| 00:01 |
kaadmy |
Malboge anyone? |
| 00:03 |
paramat |
now merging to game |
| 00:04 |
cmdskp |
happy merging! :) |
| 00:04 |
cmdskp |
we really needs some more code holiday spirit... |
| 00:08 |
paramat |
merged |
| 00:10 |
Fixer |
any news about water over ignore problem? |
| 00:10 |
red-001 |
no |
| 00:11 |
red-001 |
it's a hard problem |
| 00:11 |
red-001 |
there are easy fixes |
| 00:12 |
paramat |
ah, gregorycu, hmmmmm isn't active so work on the water-ignore thing would be appreciated if you feel like it |
| 00:12 |
red-001 |
but water is just not done well |
| 00:12 |
gregorycu |
Bug report? |
| 00:13 |
paramat |
the thing you almost started working on before |
| 00:13 |
gregorycu |
Fixed, which problem are you talking about? |
| 00:13 |
red-001 |
read this discussion https://github.com/minetest/minetest/pull/2977 |
| 00:14 |
red-001 |
and this https://github.com/minetest/minetest/issues/2767 |
| 00:14 |
gregorycu |
Ok, that one |
| 00:14 |
paramat |
yeah 2977, the method suggested by hmmmmm |
| 00:14 |
gregorycu |
Jeez, nothing has happened with that |
| 00:14 |
red-001 |
when water reaches ignore |
| 00:14 |
red-001 |
it flows over it |
| 00:14 |
gregorycu |
kwolekr's solution |
| 00:14 |
gregorycu |
Yeah |
| 00:15 |
red-001 |
well |
| 00:15 |
red-001 |
the problem is you how to up date the nodes |
| 00:15 |
sapier |
https://github.com/minetest/minetest/pull/3493 any oppinions on this one? |
| 00:15 |
red-001 |
if the user restart the server |
| 00:16 |
red-001 |
so some sort of queue that gets saved |
| 00:16 |
red-001 |
or metadata |
| 00:17 |
gregorycu |
Another solution is that when a block is loaded, it simply calls transform liquids on all liquids that are touching that block |
| 00:17 |
gregorycu |
(Rather queues them) |
| 00:17 |
red-001 |
well appertly that is too reasource hungery |
| 00:17 |
red-001 |
to PilzAdam that is |
| 00:18 |
gregorycu |
Water is faster than it used to be |
| 00:18 |
red-001 |
yeah |
| 00:18 |
red-001 |
it is |
| 00:18 |
gregorycu |
I made it faster |
| 00:18 |
gregorycu |
So maybe that gives us some wiggle room |
| 00:19 |
gregorycu |
I think what you suggested isn't exactly what I suggested |
| 00:19 |
red-001 |
I will try to do something with this when I have a chance |
| 00:19 |
red-001 |
you are welcome to do it if you want to |
| 00:19 |
gregorycu |
I am saying that when a block is loaded, queue the water touching that block, to be done by the transform liquids thread |
| 00:19 |
red-001 |
yes |
| 00:19 |
paramat |
on the subject https://github.com/minetest/minetest/pull/3501 'Liquids: Flush non walkable and buildable_to nodes away' |
| 00:20 |
gregorycu |
I think you suggested doing it in the load of the block (on the same thread) |
| 00:20 |
red-001 |
on the subject of PR If was told to show this to you |
| 00:20 |
red-001 |
https://github.com/minetest/minetest_game/pull/667 |
| 00:21 |
red-001 |
https://github.com/minetest/minetest_game/pull/602 |
| 00:21 |
red-001 |
paramat |
| 00:22 |
paramat |
yeah i've seen, you've posted the links many times :) if i don't comment it's because i'm neutral or too busy to review |
| 00:23 |
red-001 |
ok thanks |
| 00:23 |
gregorycu |
Before I switch over to this water thing, with regards to the lighting perf, I use a vector instead of a set, a vector will have bad performance if it has to reallocate lots, which may explain why a massive 112^3 chunk may be slow |
| 00:23 |
red-001 |
sometimes I wonder are you offline |
| 00:24 |
paramat |
ok most schematics are much smaller, so my test is unusual |
| 00:24 |
gregorycu |
Apparently something near that is common with mapgens |
| 00:24 |
paramat |
no mapgen uses a different faster 'spreadLight' |
| 00:25 |
paramat |
in mapgen.cpp |
| 00:25 |
gregorycu |
Well, i want to prove this is the cause |
| 00:25 |
gregorycu |
If it is, I can simply make the vector reserve the chunk size |
| 00:25 |
gregorycu |
Which means 4MB worth of vector in your case |
| 00:26 |
paramat |
cool |
| 00:26 |
red-001 |
so where does mapblock loading happen? |
| 00:26 |
gregorycu |
I know you're busy, what % of the 1.4 million blocks were lights? |
| 00:27 |
gregorycu |
Or rather, light sources |
| 00:27 |
paramat |
erm 350 of those |
| 00:27 |
gregorycu |
Not that much, really |
| 00:28 |
paramat |
i've added comments to your PR |
| 00:28 |
gregorycu |
Yeah, going through them |
| 00:28 |
paramat |
ok |
| 00:28 |
gregorycu |
red-001: EmergeThread::run |
| 00:29 |
red-001 |
thanks |
| 00:29 |
gregorycu |
Servermap::LoadBlock |
| 00:29 |
red-001 |
so thats what the server calls when I wants to load a new block into memory? |
| 00:29 |
red-001 |
it* |
| 00:30 |
gregorycu |
Well |
| 00:30 |
gregorycu |
It's the thing that does it |
| 00:30 |
gregorycu |
It's async |
| 00:30 |
red-001 |
ok |
| 00:30 |
gregorycu |
Not sure what code kicks off the process for a block |
| 00:31 |
* red-001 |
wish there was a bit more documentation |
| 00:31 |
gregorycu |
With regards to the issue, how the hell is this meant to work with oceans |
| 00:31 |
gregorycu |
Where the bottom of the ocean may not be loaded |
| 00:31 |
red-001 |
yeah |
| 00:32 |
red-001 |
thats one of the problems |
| 00:32 |
red-001 |
blow up part of an ocean with overpowerd TNT |
| 00:32 |
red-001 |
and see how bad liquid bugs get |
| 00:33 |
gregorycu |
What is an ocean? |
| 00:33 |
gregorycu |
It's whatever the mapgen decides it is |
| 00:33 |
red-001 |
:) |
| 00:33 |
gregorycu |
A mapgen can cause it to end |
| 00:33 |
cmdskp |
that sounds intriguing... |
| 00:33 |
gregorycu |
Abruptly |
| 00:33 |
gregorycu |
And form a waterfall |
| 00:33 |
gregorycu |
Deliberatly |
| 00:34 |
* red-001 |
finds overpowered tnt a *bit* too fun |
| 00:34 |
red-001 |
well that world is gone now |
| 00:35 |
red-001 |
in the name of testing |
| 00:36 |
gregorycu |
Actually, I may be wrong here |
| 00:36 |
red-001 |
yeah |
| 00:36 |
red-001 |
it doesn't look right |
| 00:37 |
red-001 |
can't find Servermap::LoadBlock |
| 00:37 |
cmdskp |
TNT-making water instead of air? ;) |
| 00:37 |
gregorycu |
ServerMap::loadBlock |
| 00:37 |
gregorycu |
line 3361 of map.cpp |
| 00:39 |
red-001 |
? |
| 00:39 |
red-001 |
nvm |
| 00:39 |
cmdskp |
abort();? |
| 00:40 |
red-001 |
the line number in my file were wrong |
| 00:40 |
gregorycu |
Sorry |
| 00:40 |
red-001 |
it wasn't minetest/master |
| 00:40 |
red-001 |
thats why |
| 00:41 |
red-001 |
not your fault |
| 00:43 |
red-001 |
there are 4 loadblock functions.... |
| 00:43 |
red-001 |
extreme overloading? |
| 00:46 |
cmdskp |
I must be out-of-date - only 3 showing up in the .h file |
| 00:46 |
gregorycu |
loadBlock(v3s16 blockpos) calls loadBlock(std::string *blob, v3s16 p3d, MapSector *sector, bool save_after_load) |
| 00:46 |
red-001 |
yeah it's 3 |
| 00:47 |
red-001 |
no 4 |
| 00:47 |
gregorycu |
I don't know where the other 2 are called |
| 00:48 |
gregorycu |
Or other 1 |
| 00:48 |
red-001 |
I have 2 more |
| 00:48 |
gregorycu |
Are you not looking at minetest/master again? |
| 00:48 |
red-001 |
yeah sorry |
| 00:48 |
red-001 |
my bad |
| 00:49 |
red-001 |
void ServerMap::loadBlock(std::string *blob, v3s16 p3d, MapSector *sector, bool save_after_load) |
| 00:51 |
|
sapier1 joined #minetest-dev |
| 00:52 |
|
sapier1 left #minetest-dev |
| 00:53 |
red-001 |
gets called https://github.com/minetest/minetest/blob/c56d7fe0eba7905b0a63c4a1cfe909988653c23d/src/map.h#L492 |
| 01:04 |
cmdskp |
@red-001: typo 'is give'->'is given,' and style-guide, I believe says a space after -- in all comments? https://github.com/red-001/minetest_game/blob/tnt-api/game_api.txt#L405 |
| 01:05 |
cmdskp |
"The radius in"->"The radius within" |
| 01:05 |
cmdskp |
oh quit =) sorry |
| 01:13 |
|
paramat left #minetest-dev |
| 01:52 |
|
Taoki_1 joined #minetest-dev |
| 01:56 |
|
Taoki joined #minetest-dev |
| 02:19 |
|
nolsen joined #minetest-dev |
| 02:19 |
gregorycu |
Fuck me more trees has significant perf issues |
| 02:19 |
|
paramat joined #minetest-dev |
| 02:20 |
gregorycu |
<gregorycu> Fuck me more trees has significant perf issues |
| 02:20 |
VanessaE |
gregorycu: aside from the mapgen lag added by biome_lib ? |
| 02:20 |
VanessaE |
the engine's spawn_tree() call has performance issues in the lighting code. |
| 02:20 |
gregorycu |
Signifiant |
| 02:21 |
gregorycu |
Damn, this is bad |
| 02:21 |
VanessaE |
how bad? |
| 02:21 |
VanessaE |
and where exactly? |
| 02:21 |
gregorycu |
33% CPU time spent in std::tree::insert |
| 02:21 |
VanessaE |
ah |
| 02:21 |
gregorycu |
(Trees are not related... lol) |
| 02:22 |
gregorycu |
I |
| 02:22 |
VanessaE |
drill down a bit and you'll probably find that it's the lighting updates that are called within the function :) |
| 02:22 |
gregorycu |
I can see it's lighting |
| 02:22 |
paramat |
gregorycu could you wait a day or 2 before working on transformLiquids? i have a commit for that i'd like to merge first |
| 02:22 |
gregorycu |
I think red is working on it |
| 02:22 |
gregorycu |
I'll post a screenshot, one moment |
| 02:23 |
gregorycu |
(for V, so she can see what I can see) |
| 02:24 |
VanessaE |
(and now I've got that damned christmas song in my head :P ) |
| 02:25 |
gregorycu |
I'll even save the screenshot as a png |
| 02:25 |
VanessaE |
heh |
| 02:26 |
paramat |
problem is l-system treegen has a lighting update for every tree, so it's meant to be used occasionally after mapgen, not multiple times during mapgen |
| 02:26 |
VanessaE |
paramat: actually, it was expressly meant to be used during mapgen |
| 02:26 |
gregorycu |
I have a feeling a can improve performance here, drastically |
| 02:26 |
VanessaE |
but your point stands. |
| 02:27 |
gregorycu |
I can improve * |
| 02:27 |
paramat |
i think l-system auto emerges 3x3x5 mapblocks per tree |
| 02:28 |
VanessaE |
paramat: you sure? I Was under the impression it was 1x3x1 |
| 02:28 |
gregorycu |
http://s27.postimg.org/x0a6dkz8z/perf_wow.png |
| 02:29 |
gregorycu |
Look at that shit |
| 02:30 |
VanessaE |
ouch |
| 02:30 |
VanessaE |
even I can read that |
| 02:30 |
gregorycu |
Inclusive samples represent the total CPU time spent doing something |
| 02:30 |
paramat |
https://github.com/minetest/minetest/blob/master/src/treegen.cpp#L124 |
| 02:30 |
paramat |
-(1,1,1) to +(1,3,1) |
| 02:30 |
gregorycu |
paramat: What does proprigateSunlight do vs spreadLight ? |
| 02:31 |
paramat |
propagates light level 15 downwards per column, it's simpler and faster than spreadlight |
| 02:31 |
gregorycu |
lol |
| 02:31 |
gregorycu |
I beg to differ |
| 02:31 |
VanessaE |
we need a whole new propagation and projection method for lighting |
| 02:32 |
paramat |
hm it should be simpler, odd |
| 02:33 |
VanessaE |
there's no reason that a shadow should be able to, for example, propagate hundreds of meters straight down. and light should be able to project farther than ~16m or whatever it is (and start out a LOT brighter) |
| 02:33 |
gregorycu |
99.5% of propagateSunlight is std::_Tree<...>::_Insert_nohint |
| 02:33 |
gregorycu |
You know what that means |
| 02:33 |
VanessaE |
in the real world, light and shadow get more and more diffuse the further they get from the source. MT needs this somehow. |
| 02:33 |
VanessaE |
gregorycu: what? |
| 02:34 |
gregorycu |
That was directed at paramat, but a similar optimisation I did to spreadLight could be done to paroagate light |
| 02:35 |
gregorycu |
Fuck my spelling today |
| 02:35 |
VanessaE |
heh. well at least you don't have to argue with auto-correct. |
| 02:37 |
paramat |
ideally treegen during mapgen would generate multiple trees in the mapchunk then do a single lighting update, if l-system treegen was meant for mapgen it is badly designed |
| 02:37 |
VanessaE |
paramat: what about caching the lighting update calls and only committing them when the destination mapblocks change? |
| 02:38 |
paramat |
currently it's only suitable for occasional saplings growing after mapgen |
| 02:38 |
VanessaE |
cache|sort|uniq --> commit at the next non-equal mapblock that is |
| 02:38 |
gregorycu |
Wasn't there an API to hold off lighting updates? |
| 02:38 |
gregorycu |
What I need to be able to do is spawn a tree on demand |
| 02:39 |
gregorycu |
A big tree |
| 02:39 |
VanessaE |
might be, but neither biome_lib nor moretrees can use it as they have no way to know which mapblocks are about to be written to once spawn_tree() is called. |
| 02:40 |
gregorycu |
ok |
| 02:40 |
gregorycu |
Well, I really want to fix this slowness |
| 02:40 |
paramat |
'spawn_ltree' should spawn a tree on demand |
| 02:40 |
paramat |
and use the giant sequoia from moretrees |
| 02:40 |
gregorycu |
Is there a way to generate a flat world? |
| 02:41 |
paramat |
.. or whatever is biggest |
| 02:41 |
paramat |
yes |
| 02:41 |
paramat |
in mgv6, 'mgv6_spflags = flat' in .conf |
| 02:42 |
gregorycu |
Thank you |
| 02:42 |
paramat |
or set that in settings |
| 02:43 |
paramat |
propagate sunlight only drops sunlight down each column until a solid node, it should be much faster than spreadLight AFAIK |
| 02:44 |
paramat |
perhaps it's broken |
| 02:45 |
gregorycu |
prop adds heaps of blocks to a std::set |
| 02:45 |
gregorycu |
Which is the slow part |
| 02:45 |
VanessaE |
in terms of tallest, I think the jungle tree might *just* squeak past the sequioa. in terms of most nodes written, that might be the oak. |
| 02:45 |
gregorycu |
As I said, apparently 99.5% of the cost of prop |
| 02:47 |
paramat |
updateLighting is known to be slow, so any improvement is very welcome |
| 02:48 |
gregorycu |
I had to make sure my other lighting fixes are good first |
| 02:49 |
gregorycu |
I have |
| 02:57 |
|
PenguinDad joined #minetest-dev |
| 03:08 |
paramat |
if l-system trees were smaller, <16 nodes in any direction, they would fit inside the mapgen's voxelmanip and only the mapgen's own lighting update would be needed. but of course they need to be larger |
| 03:12 |
paramat |
https://github.com/minetest/minetest/pull/3501 'Liquids: Flow into and destroy airlike drawtype nodes' almost ready i just need to test. bbl |
| 03:12 |
|
paramat left #minetest-dev |
| 03:18 |
gregorycu |
V: with Lua, how do I call this from my mod? moretrees:grow_jungletree(pointed_thing.above) |
| 03:18 |
gregorycu |
V = VanessaE |
| 03:19 |
VanessaE |
er |
| 03:20 |
VanessaE |
I'm not sure I understand the question |
| 03:20 |
* VanessaE |
looks at her code |
| 03:20 |
VanessaE |
actually it's more or less just that. |
| 03:21 |
VanessaE |
except use a period, not a colon. |
| 03:23 |
VanessaE |
and of course make sure you put moretrees? into your depends.txt |
| 03:23 |
gregorycu |
Yeah, I forgot that step lol |
| 03:23 |
gregorycu |
Hooray, I can spawn trees |
| 03:23 |
gregorycu |
I has the power |
| 03:23 |
VanessaE |
good deal |
| 03:23 |
VanessaE |
oh shit |
| 03:23 |
VanessaE |
he has power |
| 03:25 |
gregorycu |
Very cool |
| 03:33 |
|
nolsen joined #minetest-dev |
| 03:42 |
|
Player2 joined #minetest-dev |
| 04:00 |
|
kaadmy joined #minetest-dev |
| 04:09 |
gregorycu |
hmm... |
| 04:09 |
gregorycu |
PropagateSunlight doesn't seem to be as bad with the latest tests |
| 04:24 |
gregorycu |
Did est ever fix up the settings? |
| 04:24 |
VanessaE |
no idea. |
| 04:25 |
VanessaE |
I know, I'm real fricken' helpful huh :) |
| 04:25 |
|
paramat joined #minetest-dev |
| 04:25 |
VanessaE |
it's all HIS fault! ^^^ |
| 04:25 |
VanessaE |
:) |
| 04:27 |
paramat |
it is |
| 04:28 |
VanessaE |
just nod and say "ok", paramat :) |
| 04:28 |
VanessaE |
(really though, greg continues to investigate the lighting perf issue) |
| 05:04 |
hmmmm |
lol |
| 05:04 |
* VanessaE |
looks at hmmmm |
| 05:04 |
hmmmm |
add a dynamic hardware light to the scene for sunlight and then you don't ever need to call propogateSunlight() again |
| 05:05 |
VanessaE |
true |
| 05:05 |
hmmmm |
there's the ultimate optimization - it looks better too |
| 05:05 |
VanessaE |
but who's gonna DO it? |
| 05:05 |
hmmmm |
the fastest code is code that doesn't run |
| 05:05 |
hmmmm |
not me hehe |
| 05:05 |
hmmmm |
it's quite easy to do though |
| 05:07 |
VanessaE |
well server-side you'd still need the old lighting code wouldn't you? |
| 05:07 |
hmmmm |
for mods that rely on it, I guess so |
| 05:08 |
VanessaE |
(for that matter, client too if/when client-side Lua comes about) |
| 05:08 |
hmmmm |
ugh |
| 05:08 |
VanessaE |
is it possible to let opengl do all the work and then query the rendered results? |
| 05:09 |
VanessaE |
(assume for the moment that the server has a full opengl stack etc) |
| 05:09 |
hmmmm |
instead of holding MT back I'd rather implement light level query by casting a ray |
| 05:09 |
hmmmm |
I think so yeah |
| 05:09 |
hmmmm |
for obvious reasons that's not plausible however |
| 05:10 |
gregorycu |
Wouldn't opengl be doing the work every frame? |
| 05:10 |
hmmmm |
hmm |
| 05:10 |
gregorycu |
I know it's in hardware |
| 05:10 |
gregorycu |
At the moment the work is done in software when there are block changes |
| 05:10 |
VanessaE |
well when a mod needs a light leven somewhere, doesn't it generally have to operate on mapblocks that are not just loaded, but being displayed by someone's client? |
| 05:10 |
VanessaE |
level* |
| 05:11 |
VanessaE |
hm, strike that. |
| 05:11 |
hmmmm |
that's possible but what if that part of the scene isn't being rendered by any clients and the mod still needs it |
| 05:11 |
VanessaE |
that doesn't account for force-loaded/anchored blocks. |
| 05:11 |
hmmmm |
i kinda figured you were going to mention that |
| 05:11 |
VanessaE |
yeah, just thought about that. |
| 05:11 |
hmmmm |
the only problem with the raycasting light level query would be that you'd need to calculate radiance of all the surrounding materials in the scene to get an accurate light level other than direct sunlight or shadow |
| 05:11 |
VanessaE |
besides, you can't trust the client(s) anyway |
| 05:12 |
hmmmm |
if you can live with that limitation you should be good to go |
| 05:12 |
VanessaE |
RBA was telling me about a hardware-assisted lighting method he found that has unlimited light sources. |
| 05:13 |
hmmmm |
eh |
| 05:13 |
hmmmm |
it doesn't exist |
| 05:13 |
hmmmm |
he's probably talking about lightmaps but that has its own set of technical problems |
| 05:13 |
VanessaE |
[12-29 06:52] <RealBadAngel> https://bitbucket.org/entity/irrrenderer/commits/branch/master |
| 05:13 |
VanessaE |
(lots of discussion surrounding it; the license is compatible with MT and it does apparently build and link okay) |
| 05:15 |
hmmmm |
oh i see |
| 05:15 |
hmmmm |
the point of irrenderer is that it has deferred shading capabilities |
| 05:15 |
VanessaE |
yeah |
| 05:15 |
hmmmm |
basically I think what RBA is hinting at is that he wants to do more work in the shaders |
| 05:16 |
VanessaE |
[12-29 13:25] <RealBadAngel> http://gamedevelopment.tutsplus.com/articles/forward-rendering-vs-deferred-rendering--gamedev-12342 |
| 05:16 |
VanessaE |
hr brogght this up |
| 05:16 |
VanessaE |
brought* |
| 05:16 |
VanessaE |
oh G*d I can't type |
| 05:16 |
hmmmm |
he wants to have everything fully lit and then cast shadows afterwards |
| 05:16 |
VanessaE |
well, to be perfectly fair, this IS what shaders were meant for. |
| 05:16 |
hmmmm |
eh |
| 05:16 |
VanessaE |
though casting shadows and other such things is probably too much right now |
| 05:17 |
hmmmm |
well if he can do it and in a non-hacky manner, that's great |
| 05:17 |
hmmmm |
i think minetest would benefit more if we were to switch to a more modern graphics engine instead of some custom hacked up irrlicht |
| 05:18 |
VanessaE |
others tend to agree |
| 05:18 |
hmmmm |
shadow casting isn't going to help do what we want to do though |
| 05:18 |
VanessaE |
which reminds me, Zeno is back |
| 05:18 |
hmmmm |
the ultimate end game is to have basically infinite dynamic full color lights |
| 05:18 |
VanessaE |
not sure if you saw |
| 05:18 |
hmmmm |
no I didn't see |
| 05:19 |
hmmmm |
right anyway |
| 05:19 |
hmmmm |
I feel like a good middle ground for now would be to come up with an algorithm to determine which light sources are the most significant in a given scene, and then trim out any of them that can't be rendered |
| 05:19 |
hmmmm |
and just use hardware dynamic lights |
| 05:20 |
hmmmm |
that would give us full RGB hardware lighting with a minimal amount of code |
| 05:20 |
VanessaE |
that's not the ones that are limited to 8 lights is it? |
| 05:20 |
hmmmm |
different cards have different limits |
| 05:20 |
hmmmm |
it is though |
| 05:20 |
VanessaE |
8 is some opengl limit |
| 05:20 |
hmmmm |
oh that stinks |
| 05:20 |
VanessaE |
nevertheless, that'll fail hard, sorry to say. |
| 05:21 |
hmmmm |
why..? |
| 05:21 |
VanessaE |
I think that sort of lighting was meant for e.g. sun and moon |
| 05:21 |
hmmmm |
it's meant to be used sparingly |
| 05:21 |
VanessaE |
because in an average Minetest world, there are often hundreds of light sources on screen. |
| 05:21 |
hmmmm |
FWIW many games use the strategy I mentioned |
| 05:21 |
hmmmm |
of course there are hundreds, but not all of them contribute significantly whatsoever to the scene |
| 05:22 |
VanessaE |
well that's kinda iffy really |
| 05:22 |
VanessaE |
imagine a street running into the distance, lights every 10 meters |
| 05:23 |
VanessaE |
with a moderate view range of 100 meters, you've already got 10-20 sources visible (depending on the construction of the street lights) |
| 05:28 |
hmmmm |
it'd probably be the closest 8 or so |
| 05:28 |
hmmmm |
again, all i'm proposing is a simple partial solution to our lighting |
| 05:28 |
VanessaE |
just doing it for the sun and moon would greatly help I'm sure |
| 05:29 |
hmmmm |
what RBA wants to do will work if it's grealy simplified |
| 05:29 |
hmmmm |
but if it's to be done the correct way, he'd essentially need to solve the radiosity problem for N light sources 60 times a second |
| 05:29 |
hmmmm |
that's hard |
| 05:29 |
VanessaE |
yeah |
| 05:34 |
VanessaE |
it's weird, as our discussion started from that whole issue surrounding "extruded" plantlike |
| 05:35 |
VanessaE |
he tried my/stu's "just stack up a bunch of 2d layers" idea; it works to some degree but isn't good enough in his mind :) |
| 05:41 |
* sofar |
walks away from the computer for a bit |
| 05:49 |
|
paramat left #minetest-dev |
| 06:36 |
|
Hunterz joined #minetest-dev |
| 07:51 |
|
Krock joined #minetest-dev |
| 07:52 |
gregorycu |
What is G*d ? |
| 07:55 |
gregorycu |
Never mind, I just googled it |
| 07:56 |
gregorycu |
Just say "God". If that upsets people, they are not adjusted well. If that upsets God, he is not adjusted well. |
| 08:01 |
VanessaE |
gregorycu: it's a Jewish custom of obscuring the name so as not to give others the opportunity to deface it. |
| 08:02 |
VanessaE |
I spell it G*d instead of the usual G-d because some people mistake the latter for goddamn. |
| 08:02 |
gregorycu |
But I can just say "Fuck God" |
| 08:02 |
VanessaE |
it's a holdover from written communications. kinda hard to vandalize a name online :P |
| 08:03 |
gregorycu |
If God gets pissy at you because someone vandalized something you wrote, he is a douche |
| 08:03 |
gregorycu |
I gotta run, new years festivities and all that |
| 08:03 |
VanessaE |
anyway offtopic for here |
| 08:04 |
* gregorycu |
& |
| 08:04 |
VanessaE |
nohup gregorycu |
| 08:04 |
VanessaE |
:) |
| 08:20 |
|
alket joined #minetest-dev |
| 09:39 |
|
Icedream joined #minetest-dev |
| 09:53 |
|
Calinou joined #minetest-dev |
| 10:18 |
|
Amaz joined #minetest-dev |
| 10:36 |
|
amadin joined #minetest-dev |
| 10:57 |
|
AnotherBrick joined #minetest-dev |
| 11:01 |
|
red-001 joined #minetest-dev |
| 11:30 |
|
Krock joined #minetest-dev |
| 11:32 |
|
rubenwardy joined #minetest-dev |
| 11:32 |
rubenwardy |
Lol: https://forum.minetest.net/viewtopic.php?f=14&t=13821&view=unread#unread |
| 11:37 |
|
RealBadAngel joined #minetest-dev |
| 11:37 |
RealBadAngel |
hi |
| 11:37 |
VanessaE |
hey RBA |
| 11:38 |
RealBadAngel |
deffered rendering (lighting) IS about having unlimited number of lightsources |
| 11:38 |
RealBadAngel |
not about shadows at all |
| 11:38 |
VanessaE |
er, yeah. we covered that. |
| 11:38 |
RealBadAngel |
in fact it cant do any shadows |
| 11:38 |
VanessaE |
but whatever it's for, it has to *work* first :) |
| 11:38 |
VanessaE |
(read: compile it into MT and put it into use) |
| 11:39 |
RealBadAngel |
also, thats impossible to use opengl renderer to calculate lightlevel as we used to it |
| 11:41 |
RealBadAngel |
i m done with irrPP right now, playing with postprocessing, will get into renderer after that |
| 11:43 |
RealBadAngel |
effects by now are promising, i do have working blur, fxaa,cel and bloom |
| 11:43 |
VanessaE |
but what about basic lighting? :-/ |
| 11:44 |
RealBadAngel |
i will just start with turning off our vertex lighting, add one dynamic sun lightsource |
| 11:45 |
VanessaE |
how will that play alongside other light sources like torches? |
| 11:45 |
RealBadAngel |
torches are gonna be added with lightmap |
| 11:45 |
RealBadAngel |
or whatever is the way with that renderer |
| 11:45 |
VanessaE |
ok |
| 11:52 |
red-001 |
rubenwardy |
| 11:52 |
red-001 |
100 usd to build minetest |
| 11:52 |
red-001 |
thats not enough |
| 12:16 |
|
Calinou joined #minetest-dev |
| 13:02 |
|
zupoman joined #minetest-dev |
| 13:29 |
|
realbadangel__ joined #minetest-dev |
| 13:52 |
red-001 |
what does fgettext() do? |
| 13:52 |
red-001 |
is it something to do with translation ? |
| 13:52 |
|
Fixer joined #minetest-dev |
| 14:21 |
|
behalebabo joined #minetest-dev |
| 14:25 |
|
Hunterz joined #minetest-dev |
| 14:27 |
|
CraigyDavi joined #minetest-dev |
| 14:48 |
|
VargaD joined #minetest-dev |
| 14:59 |
|
rubenwardy joined #minetest-dev |
| 15:22 |
|
kaadmy joined #minetest-dev |
| 15:36 |
realbadangel__ |
red-001, yeah |
| 15:38 |
realbadangel__ |
if theres no locale you will get original, en, otherwise translated strings |
| 15:39 |
red-001 |
t |
| 15:39 |
red-001 |
thnaks |
| 15:41 |
red-001 |
I noticed a strange bug when I place 5 nodes like this http://imgur.com/yzRKhFc I can climb to the top node |
| 15:42 |
red-001 |
Is this a known bug? |
| 15:43 |
RealBadAngel |
going straight forward and jump? |
| 15:43 |
red-001 |
no I held shift |
| 15:44 |
kaadmy |
not surprised |
| 15:44 |
kaadmy |
sneakjumping is very glitchy |
| 15:45 |
RealBadAngel |
AWWW |
| 15:45 |
RealBadAngel |
this is insane |
| 15:46 |
RealBadAngel |
sneak gives a rocket in the ass when jumpin |
| 15:46 |
RealBadAngel |
who the fuck did that? |
| 15:47 |
RealBadAngel |
sneak (crouch) should eliminate ability to jump at the same time |
| 15:47 |
RealBadAngel |
who the fuck did that logic?? |
| 15:48 |
red-001 |
is it called sneak jumping and a setting |
| 15:48 |
red-001 |
I thought It was disabled by default |
| 15:51 |
|
proller joined #minetest-dev |
| 15:52 |
RealBadAngel |
imho such behaviour should not be possible at all |
| 15:52 |
RealBadAngel |
this is not quake, rocket jumping should not work here |
| 15:52 |
red-001 |
well it sure cause some bugs |
| 15:53 |
RealBadAngel |
unless theres another player standing behind you and kick you in the ass ;) |
| 15:54 |
Calinou |
<+RealBadAngel> sneak (crouch) should eliminate ability to jump at the same time |
| 15:54 |
Calinou |
sneaking is not crouching |
| 15:54 |
Calinou |
it's akin to walking in a FPS |
| 15:55 |
Calinou |
also, in Quake you can't crouch and jump at the same time, attempting to do so will just uncrouch you |
| 15:55 |
Calinou |
this does not apply to Xonotic though |
| 15:55 |
Calinou |
(also, Quake 1 had no crouching. only walk/run) |
| 15:56 |
red-001 |
in source engine games you jump higher if you crouch in the aor |
| 15:56 |
red-001 |
air* |
| 15:56 |
RealBadAngel |
Calinou, ok, youre right, but look at this again |
| 15:57 |
Calinou |
red-001, exactly, crouching in those raises your legs in the air |
| 15:57 |
gravgun |
You don't jump higher tho |
| 15:57 |
RealBadAngel |
when youre walking slowly, you jump further irl? |
| 15:57 |
Calinou |
makes more sense than lowering head (which is what Quake does when you jump then crouch) |
| 15:57 |
Calinou |
(Quake 3 allowed crouching while in air, Quake 2 doesn't) |
| 15:57 |
Calinou |
RealBadAngel, no you don't |
| 15:58 |
Calinou |
sneaking could make jumping not as high maybe |
| 15:58 |
Calinou |
(ideally, should be configurable) |
| 15:58 |
RealBadAngel |
not at all |
| 15:59 |
RealBadAngel |
making everything configureable is leading us to the same place as keeping eternal backwards compability |
| 15:59 |
Calinou |
hardcoding is bad m'kay? |
| 15:59 |
RealBadAngel |
clusterfuck |
| 15:59 |
RealBadAngel |
clusterm'kay |
| 16:00 |
RealBadAngel |
imho some things have to be hardcoded |
| 16:00 |
Calinou |
I don't think so, we're supposed to be an extensible engine |
| 16:00 |
red-001 |
at some point reading minetest.conf,example will be as hard a complying form source |
| 16:00 |
RealBadAngel |
we cant make full, realistic phys engine |
| 16:01 |
RealBadAngel |
it already is |
| 16:01 |
red-001 |
well not on windows |
| 16:01 |
red-001 |
on linux yes |
| 16:01 |
RealBadAngel |
even other devs are claimin that some things "are barely useable, becasue not on default" |
| 16:02 |
RealBadAngel |
and theyre right |
| 16:02 |
Calinou |
why don't we use multiple configuration files? or a configuration file with sections? |
| 16:02 |
red-001 |
like source engine config files? |
| 16:03 |
Calinou |
like PHP, nginx… |
| 16:03 |
Calinou |
we can have a configuration file 1) for client, 2) for server, 3) for common settings |
| 16:03 |
Calinou |
most games split by type like this already |
| 16:03 |
RealBadAngel |
game shall have settings like this: basic, gimme some more, ouch, hardcore |
| 16:04 |
red-001 |
and death |
| 16:04 |
RealBadAngel |
without users being able to set fsaa sampling rate |
| 16:04 |
kaadmy |
RBA: good idea ;) |
| 16:04 |
RealBadAngel |
and it should be up to us to decide what we can enable |
| 16:04 |
RealBadAngel |
based maybe on some fast client test |
| 16:05 |
red-001 |
isn't that the case now? |
| 16:05 |
RealBadAngel |
that shouldnt be any hard to check for whats aviable and how fast the client box is |
| 16:05 |
red-001 |
I doubt most users check minetest.conf |
| 16:06 |
RealBadAngel |
most of the users cannot edit iot |
| 16:06 |
RealBadAngel |
*it |
| 16:06 |
RealBadAngel |
thats the real problem |
| 16:06 |
red-001 |
what is the use of the mod tab |
| 16:07 |
red-001 |
all it does is list what is in /mods |
| 16:07 |
RealBadAngel |
we are the elite, something whats obvious for us, is way beyond imagination of typical gamer |
| 16:07 |
red-001 |
minetest master race? |
| 16:07 |
RealBadAngel |
and this is our problem we cant see it |
| 16:07 |
RealBadAngel |
no |
| 16:08 |
RealBadAngel |
we are coders |
| 16:08 |
red-001 |
true |
| 16:08 |
RealBadAngel |
players are not |
| 16:08 |
red-001 |
!false |
| 16:09 |
RealBadAngel |
within my friends, family, customers i used to run google and keep looking for solutions to their problems |
| 16:10 |
RealBadAngel |
regular user is dumb enough to not know how to search google even |
| 16:10 |
RealBadAngel |
how could you expect him to edit config files? |
| 16:10 |
red-001 |
or what to search for |
| 16:10 |
red-001 |
you can |
| 16:11 |
red-001 |
can't* |
| 16:11 |
red-001 |
or use the new settings tab |
| 16:11 |
RealBadAngel |
atm our course is developing a product for same freaks as we are |
| 16:12 |
RealBadAngel |
eliminating kids, unexperienced or just plain dumb users at the start |
| 16:12 |
red-001 |
and not all of us even play it a lot |
| 16:12 |
RealBadAngel |
i do |
| 16:13 |
RealBadAngel |
im not typical core dev and you know it |
| 16:13 |
red-001 |
yeah I know |
| 16:13 |
RealBadAngel |
i code for my own gamin pleasure |
| 16:13 |
red-001 |
that's why we have awesome things like the minimap? |
| 16:14 |
RealBadAngel |
brb, last time to buy something before new year ;) |
| 16:25 |
|
rubenwardy joined #minetest-dev |
| 16:31 |
|
zat joined #minetest-dev |
| 16:34 |
|
Gael-de-Sailly joined #minetest-dev |
| 16:37 |
RealBadAngel |
red-001, im not coding stuff that i cannot see use in game for |
| 16:38 |
|
Fixer joined #minetest-dev |
| 16:40 |
red-001 |
is bultin\settingtypes.txt where settings have to be added to be seen on the settings page? |
| 16:41 |
RealBadAngel |
yes |
| 17:00 |
|
cmdskp joined #minetest-dev |
| 17:21 |
cmdskp |
'Slip'ping on with: https://github.com/minetest/minetest/pull/3506 |
| 17:27 |
red-001 |
does minetest have a lua profiler ? |
| 17:35 |
|
ElectronLibre joined #minetest-dev |
| 17:36 |
|
hmmmm joined #minetest-dev |
| 17:53 |
|
celeron55_ joined #minetest-dev |
| 17:58 |
Krock |
red-001, yes, there's profiler tool ni the builtin code |
| 17:58 |
red-001 |
how to use it? |
| 17:58 |
Krock |
builtin/game/mod_profiling.lua |
| 17:59 |
red-001 |
ok thanks |
| 17:59 |
Krock |
profiling_print_log("", "") |
| 17:59 |
Krock |
outputs stats |
| 17:59 |
Krock |
alright, there's also a minetest.conf setting: detailed_profiling |
| 18:03 |
|
celeron55 joined #minetest-dev |
| 18:06 |
|
Darcidride joined #minetest-dev |
| 18:06 |
|
Darcidride_ joined #minetest-dev |
| 18:13 |
|
celeron55 joined #minetest-dev |
| 18:25 |
|
celeron55 joined #minetest-dev |
| 18:27 |
|
Drangue joined #minetest-dev |
| 18:36 |
|
celeron55 joined #minetest-dev |
| 18:45 |
|
celeron55 joined #minetest-dev |
| 18:46 |
|
ElectronLibre joined #minetest-dev |
| 18:56 |
|
celeron55 joined #minetest-dev |
| 18:57 |
|
red-001 joined #minetest-dev |
| 19:07 |
|
celeron55 joined #minetest-dev |
| 19:17 |
|
celeron55 joined #minetest-dev |
| 19:29 |
|
celeron55 joined #minetest-dev |
| 19:42 |
|
nolsen joined #minetest-dev |
| 19:46 |
|
Wayward1 joined #minetest-dev |
| 19:47 |
Wayward1 |
not sure if it's my fault or not, but lately all my builds are failing with "No rule to make target `/usr/lib/libGL.so', needed by `bin/minetest'." |
| 19:47 |
Wayward1 |
any ideas? |
| 19:48 |
red-001 |
you need opengl? |
| 19:48 |
red-001 |
what os do you have? |
| 19:49 |
cmdskp |
sounds like the path is wrong to libGL in the make |
| 19:51 |
red-001 |
the mod tab in main menu is useless |
| 19:51 |
Wayward1 |
i've had all the dependencies installed for a while. come to think of it, i did update a few recently... |
| 19:52 |
Wayward1 |
linux mint 17.2 btw |
| 19:52 |
cmdskp |
probably the updates have moved the lib into a different (sub)folder, perhaps? |
| 19:53 |
Wayward1 |
hmm, could be. i'll check |
| 19:55 |
red-001 |
You can't even delate mods form it |
| 19:55 |
cmdskp |
it's only for turning them(or parts of them) on/off |
| 19:55 |
cmdskp |
It would be nice to be able to delete |
| 19:55 |
red-001 |
the mods tab? |
| 19:55 |
red-001 |
it's in the main menu |
| 19:56 |
red-001 |
at the top left beside credits |
| 19:56 |
cmdskp |
oh, I see it seems bugged |
| 19:56 |
cmdskp |
I usually use the Configure button to access the mods |
| 19:56 |
Wayward1 |
cmdskp: that was it, thanks. found it in a system-wide search |
| 19:56 |
red-001 |
it just lists mods and there depends and the delate button doesn't work |
| 19:57 |
red-001 |
cmdskp |
| 19:57 |
red-001 |
so do I |
| 19:57 |
cmdskp |
Oh, I get it! It shows a description and screenshot as well |
| 19:58 |
red-001 |
imo it should get merged with the configure opitions |
| 19:58 |
cmdskp |
It could become more useful if it looked up a repository of mods and listed them... |
| 19:58 |
red-001 |
It used to |
| 19:58 |
cmdskp |
similar to how servers tab works |
| 19:58 |
cmdskp |
ah |
| 19:58 |
red-001 |
the database is gone |
| 19:58 |
cmdskp |
that's a pity |
| 19:58 |
red-001 |
yeah |
| 19:59 |
red-001 |
It looks better then the mod configure options |
| 20:00 |
red-001 |
It could be used to replace them |
| 20:00 |
|
Icedream joined #minetest-dev |
| 20:08 |
|
zupomanek joined #minetest-dev |
| 20:15 |
|
Gael-de-Sailly joined #minetest-dev |
| 20:21 |
|
Wayward1 joined #minetest-dev |
| 21:00 |
|
alket joined #minetest-dev |
| 21:03 |
|
cmdskp joined #minetest-dev |
| 21:04 |
|
Drangue joined #minetest-dev |
| 21:17 |
|
ElectronLibre joined #minetest-dev |
| 21:40 |
|
rubenwardy joined #minetest-dev |
| 21:55 |
cmdskp |
Is it normal when runnning minetest to have a stream in Visual Studio Output(but not the console window) of exception: SettingNotFoundException at memory location 0x...? |
| 22:27 |
|
celeron55 joined #minetest-dev |
| 22:27 |
|
rubenwardy_ joined #minetest-dev |
| 22:28 |
|
Warr1024_ joined #minetest-dev |
| 22:53 |
RealBadAngel |
Happy New year! |
| 22:53 |
|
Warr1024 joined #minetest-dev |
| 22:57 |
|
paramat joined #minetest-dev |
| 22:59 |
|
paramat left #minetest-dev |
| 23:00 |
kaadmy |
happy new year! |
| 23:07 |
red-001 |
It's an hour away here |
| 23:15 |
kaadmy |
about 9 hours for me ;) |
| 23:23 |
red-001 |
what timezone do you live in? |
| 23:25 |
kaadmy |
red-001: UTC-8 |
| 23:26 |
* red-001 |
thought UTC-8 was UTC+8 |
| 23:35 |
red-001 |
there is a bug that causes minetest to lock the cursor to the center of the screen after loading the client even if the minetest window is not active |
| 23:35 |
red-001 |
does this happen to anyone else? |
| 23:36 |
red-001 |
there is no way to unlock the cursor |
| 23:36 |
red-001 |
but logout |
| 23:37 |
rubenwardy |
That happens to me sometimes, but I can pause the game and free the cursor? |
| 23:38 |
red-001 |
I can't |
| 23:38 |
red-001 |
not usually |
| 23:39 |
rubenwardy |
What OS do you use? |
| 23:39 |
red-001 |
windows 7 |
| 23:40 |
red-001 |
are you on linux? |
| 23:40 |
kaadmy |
red-001: doesn't happen for me, but I use Linux. I just Ctrl+Alt+F1 and type `kill minetest` :) |
| 23:40 |
kaadmy |
after i log in, of course ;) |
| 23:40 |
rubenwardy |
Ah. I use the XFCE dekstop environment and ubuntu. So xubuntu |
| 23:40 |
rubenwardy |
Linux |
| 23:40 |
red-001 |
yeah I rember it happening a few times on linux too |
| 23:41 |
red-001 |
at least 'esc' worked there |
| 23:41 |
kaadmy |
i also use i3, so I can also just do Mod+Q |
| 23:41 |
kaadmy |
actually Mod+Shift+Q |
| 23:42 |
red-001 |
what does that do? |
| 23:42 |
red-001 |
mod as in 'alt' key? |
| 23:42 |
kaadmy |
my mod key is currently the Super/Windows key closes the currently selected window ;) |
| 23:42 |
red-001 |
ahh |
| 23:43 |
kaadmy |
with my window manager only tho.. |
| 23:43 |
kaadmy |
on windows, does Alt+F4(i think) work? |
| 23:43 |
red-001 |
well I have the multiple virtual desktops so it's ok for me |
| 23:43 |
red-001 |
but I don't think a lot of people have that on windows |
| 23:43 |
kaadmy |
i have keybindings to do almost everything, so the mouse being captured in an invisible window is trivial ;) |
| 23:44 |
kaadmy |
or i can run `xkill` |
| 23:44 |
red-001 |
of course |
| 23:47 |
red-001 |
well I think I found a way to reproduce it |
| 23:48 |
red-001 |
start minetest let it start loading and then leave the window maximized and open an other window |
| 23:51 |
kaadmy |
nope, that doesn't happen for me |
| 23:54 |
red-001 |
well what distro do you use? |
| 23:55 |
kaadmy |
fork of Debian called Kali |
| 23:55 |
red-001 |
It happend to me a few times in Mint a ubuntu based distro |
| 23:55 |
kaadmy |
with i3 as a WM, Gnome might have the problem though |
| 23:56 |
red-001 |
mint uses a clone if Gnome |
| 23:56 |
red-001 |
cinnamon |
| 23:58 |
red-001 |
happy new year |
| 23:59 |
red-001 |
to everyone in GMT |
