| Time |
Nick |
Message |
| 00:00 |
iqualfragile |
sapier: sounds incredibly cpu killing |
| 00:00 |
hmmmm |
Minecraft's method is to make the terrain based on the biomes, which only works because the whole terrain is interpolated to begin with |
| 00:00 |
NakedFury |
could it be possible to have some terraingen operations done in c++ in some sort of prepackage so when you in lua call them they calculate faster than would be possible in lua? |
| 00:00 |
celeron55 |
sapier: that doesn't make sense and is a thousand times more heavy |
| 00:00 |
hmmmm |
I'm explicitly trying to avoid that |
| 00:01 |
sapier |
I know it sounds a lot more heavy ... and in total it will be but on the other hand you can do it incremently |
| 00:01 |
celeron55 |
the map is generated at runtime anyway; it's not feasible for it to be more runtimey |
| 00:01 |
iqualfragile |
sapier i do not think that would work out |
| 00:02 |
sapier |
ok it's just been an idea |
| 00:02 |
hmmmm |
so |
| 00:02 |
hmmmm |
I don't want to make the biome system shape the way mapgens work |
| 00:02 |
hmmmm |
I wanted the biome system to be able to be placed on top of any mapgen |
| 00:03 |
celeron55 |
(i can agree that would be ideal) |
| 00:03 |
hmmmm |
if we start making some concoction of the noise used to generate the terrain into the actual terrain variance values, and it works, would we really want that system? |
| 00:03 |
hmmmm |
keep in mind that mapgen v7 is not just 2d perlin, but this mishmash of different processes |
| 00:04 |
hmmmm |
also this doesn't help with finding the difference between say, hills and ocean (where absolute ground height matters and not the variance) |
| 00:04 |
celeron55 |
the issue is that doing nonlinear algorithms in chunks while a continous output is required is impossible |
| 00:05 |
hmmmm |
i'm trying to find a way to approximate the process instead of doing something impossible |
| 00:05 |
hmmmm |
it's just really hard |
| 00:05 |
hmmmm |
this is why mapgen v7 is taking so long |
| 00:06 |
hmmmm |
the reason why biomes just werk in minecraft is because it's completely bound to one mapgen - the mapgen - and there's 0 flexibility |
| 00:06 |
hmmmm |
i will not tolerate such a crude approach |
| 00:06 |
celeron55 |
so either you really only have stuff like gaussian blur at your disposal (even that becomes a pain when some but not all of the surrounding terrain is generated) or you do whatever you want and tweak it until it tends to blend together |
| 00:07 |
hmmmm |
celeron, I'm thinking to combat that it might be a good idea to just overgenerate the raw terrain (in the x, z directions, not all 3) |
| 00:07 |
celeron55 |
personally i will hugely prefer a thing like you are discussing; i just know it's not trivial at all |
| 00:07 |
hmmmm |
that has its own set of problems too |
| 00:08 |
iqualfragile |
celeron55: like split terrain and surface generation? |
| 00:10 |
celeron55 |
also, it has to be able to work in some reasonable way with eg. overhangs and flying islands |
| 00:10 |
iqualfragile |
overhanging cliffs would be a challenge |
| 00:10 |
iqualfragile |
flying islands could be generated later on |
| 00:11 |
celeron55 |
sounds to me like multiple passes of some kind of preprocessing or something maybe |
| 00:11 |
iqualfragile |
yep |
| 00:12 |
celeron55 |
or whatever, it's actually dumb to guess now because it depends on how it will actually even work for basic cases |
| 00:14 |
iqualfragile |
well, there would definitly be at least two steps, terrain and biomes |
| 00:14 |
celeron55 |
...eh what, we are discussing solely about the biome part by now |
| 00:15 |
iqualfragile |
in that case i missunderstood something |
| 00:16 |
celeron55 |
hmmmm: there are also weird options like using a GPU to do computation; it allows doing certain things that are totally infeasible on CPU |
| 00:17 |
celeron55 |
like applying a complex filter on a bitmap |
| 00:17 |
hmmmm |
all for something as mundane as choosing biomes |
| 00:17 |
hmmmm |
this keeps getting better and better |
| 00:17 |
celeron55 |
well it probably isn't something we want to do |
| 00:18 |
hmmmm |
i think a lot of people in the minetest community don't appreciate that this is a difficult problem |
| 00:18 |
celeron55 |
but if someone wants to try things for science, then that's an option |
| 00:19 |
celeron55 |
a lot of people don't appreciate anything; that's not anything new 8) |
| 00:19 |
iqualfragile |
celeron55: problem would be that gpu acceleration is not avivable in every case |
| 00:19 |
celeron55 |
iqualfragile: that is the obvious reason why < celeron55> well it probably isn't something we want to do |
| 00:20 |
iqualfragile |
jup |
| 00:21 |
celeron55 |
let's implement minetest's mapgen on an FPGA PCIe card |
| 00:21 |
celeron55 |
practically nobody could use it but we'd get in the news! |
| 00:23 |
celeron55 |
hmmmm: anyway, so, do you happen to have time for actually doing it or is this again just one of the regular "this would be cool except nobody will ever want to spend the time to make it"? 8) |
| 00:24 |
celeron55 |
i'm assuming the latter |
| 00:25 |
|
sapier1 joined #minetest-dev |
| 00:29 |
hmmmm |
well I have wanted to get the biome thing done for quite a long time |
| 00:29 |
hmmmm |
if I can come up with something that actually works well, then yeah.. |
| 00:29 |
hmmmm |
it'd be very motivational |
| 00:30 |
celeron55 |
we might be in a lack of a usable system for testing such things |
| 00:30 |
celeron55 |
firing up regular minetest for it is kind of clunky |
| 00:30 |
hmmmm |
i made a small GTK+ application for testing mapgen related crap fast |
| 00:31 |
celeron55 |
is it public? i guess more people could have use for it |
| 00:31 |
hmmmm |
it's not because it's just a testing framework |
| 00:31 |
hmmmm |
very rough around the edges |
| 00:52 |
paramat |
i would like to suggest the future minetest underworld realm be named after a Finnish mythological underworld, im fed up with the n***** word |
| 00:55 |
paramat |
also, spring equinox is soon, 0.4.9 is 3 months old and was broken (im using 0.4.8) perhaps time for 0.4.10? |
| 00:56 |
hmmmm |
soon(tm) |
| 01:01 |
|
OldCoder joined #minetest-dev |
| 01:24 |
ShadowNinja |
The current version of Minetest seems fairly stable. But I'd suggest that next time we decide to make a release we wait at least 12 hours after announcing it before releasing. |
| 01:29 |
iqualfragile |
:D |
| 01:31 |
sapier1 |
can someone explain to me why we use https for modstore list but send password hash unencrypted? |
| 01:32 |
iqualfragile |
sapier1: well, the certificate for the modstore is signed by some ca |
| 01:33 |
sapier1 |
doesn't help without ssl support in android curl version (yet) |
| 01:36 |
|
Megaf joined #minetest-dev |
| 01:49 |
|
sapier1 left #minetest-dev |
| 01:54 |
|
Megaf_ joined #minetest-dev |
| 01:54 |
|
werwerwer joined #minetest-dev |
| 02:05 |
ShadowNinja |
SSL support has been discussed before. sfan5: Do you think that you have the ability to properly implement it? It would have to be negotiated by a pre-init message. |
| 04:19 |
|
proller joined #minetest-dev |
| 04:45 |
|
NakedFury joined #minetest-dev |
| 04:53 |
|
khonkhortisan joined #minetest-dev |
| 06:20 |
|
darkrose joined #minetest-dev |
| 06:20 |
|
darkrose joined #minetest-dev |
| 06:43 |
sfan5 |
ShadowNinja: you mean TLS, right? |
| 06:44 |
|
ImQ009 joined #minetest-dev |
| 06:44 |
|
troller joined #minetest-dev |
| 06:46 |
ShadowNinja |
sfan5: Yes, TLSv1.2 AES 256 with ECDHE and SHA-512. |
| 06:47 |
sfan5 |
to make stuff secure I would go for a solid default config |
| 06:53 |
sfan5 |
which TLS library shall we use? |
| 06:54 |
sfan5 |
GnuTLS, OpenSSL? |
| 06:54 |
sfan5 |
s/,/?/ |
| 06:57 |
ShadowNinja |
sfan5: GNUTLS would be better, OpenSSL has licemsing issues IIRC. |
| 06:58 |
sfan5 |
and gnutls seems to support a ton more while being smaller than openssl |
| 06:58 |
sfan5 |
http://en.wikipedia.org/wiki/Comparison_of_TLS_implementations#Code_size_and_dependencies |
| 07:02 |
sfan5 |
We won't be doing anything with signed certificates, right? |
| 07:03 |
sfan5 |
(I mean having a CA sign them) |
| 07:10 |
sfan5 |
will we use TLS over connection.cpp's system of packets or will we use DTLS |
| 07:10 |
sfan5 |
? |
| 07:16 |
|
nore joined #minetest-dev |
| 07:36 |
|
khonkhortisan joined #minetest-dev |
| 07:59 |
|
PenguinDad joined #minetest-dev |
| 08:09 |
|
Calinou joined #minetest-dev |
| 08:18 |
|
Hiradur joined #minetest-dev |
| 08:34 |
|
alexxs joined #minetest-dev |
| 09:22 |
|
restcoser joined #minetest-dev |
| 09:31 |
smoke_fumus |
ok guys i really suck at trigonometry and i need a little help down here. i have a view (2d screen) and a 3d space around it. i need to find a position on screen at which object i am looking at located and get screen coordinate out of it; object appears on screen and i need to check if it is in field of view and then retrieve screen coordinate |
| 09:43 |
celeron55 |
use ISceneCollisionManager |
| 09:43 |
celeron55 |
http://irrlicht.sourceforge.net/docu/classirr_1_1scene_1_1_i_scene_collision_manager.html#a6032377ff769e42c3e28547794f015ea |
| 09:44 |
celeron55 |
no need to know trigonometry 8) |
| 09:44 |
celeron55 |
of course if this isn't in minetest then you may or may not have a similar one available |
| 09:45 |
celeron55 |
in that case this is what it does http://paste.ubuntu.com/7100846/ |
| 09:49 |
celeron55 |
(which is, it just multiplies the position with the view and projection matrix) |
| 10:31 |
|
rsiska joined #minetest-dev |
| 10:39 |
|
Jordach joined #minetest-dev |
| 10:49 |
|
Calinou joined #minetest-dev |
| 10:50 |
|
restcoser joined #minetest-dev |
| 11:02 |
realbadangel_ |
smoke_fumus, such thing is already done in hud code |
| 11:02 |
smoke_fumus |
realbadangel_: nonono, not for minetest |
| 11:02 |
smoke_fumus |
aside from minetest |
| 11:03 |
realbadangel_ |
you can still read the code |
| 11:05 |
|
restcoser joined #minetest-dev |
| 11:59 |
|
PilzAdam joined #minetest-dev |
| 12:02 |
|
iqualfragile joined #minetest-dev |
| 12:16 |
|
rsiska joined #minetest-dev |
| 12:17 |
|
grrk-bzzt joined #minetest-dev |
| 12:48 |
|
PenguinDad joined #minetest-dev |
| 13:07 |
|
alexxs joined #minetest-dev |
| 13:10 |
|
deltib joined #minetest-dev |
| 13:21 |
|
PenguinDad joined #minetest-dev |
| 13:32 |
|
Zeitgeist_ joined #minetest-dev |
| 13:32 |
|
Zeitgeist_ joined #minetest-dev |
| 13:40 |
|
NakedFury joined #minetest-dev |
| 14:04 |
|
Jordach_ joined #minetest-dev |
| 14:17 |
|
grrk-bzzt joined #minetest-dev |
| 14:21 |
|
tomreyn joined #minetest-dev |
| 14:39 |
|
sapier joined #minetest-dev |
| 14:47 |
|
Zeitgeist_ joined #minetest-dev |
| 14:47 |
|
Zeitgeist_ joined #minetest-dev |
| 14:49 |
|
Zeitgeist_ joined #minetest-dev |
| 14:56 |
|
Zeitgeist_ joined #minetest-dev |
| 14:56 |
|
Zeitgeist_ joined #minetest-dev |
| 15:02 |
|
hmmmm joined #minetest-dev |
| 15:06 |
|
Zeitgeist_ joined #minetest-dev |
| 15:06 |
|
Zeitgeist_ joined #minetest-dev |
| 15:11 |
|
alexxs joined #minetest-dev |
| 15:13 |
|
Ritchie joined #minetest-dev |
| 15:36 |
|
Ritchie joined #minetest-dev |
| 15:53 |
|
iqualfragile_ joined #minetest-dev |
| 16:01 |
|
rdococ joined #minetest-dev |
| 16:01 |
rdococ |
Hello dear developers. |
| 16:02 |
ShadowNinja |
Hello rdococ. What is your query? |
| 16:02 |
rdococ |
I can not for the life of me find the mod API's name for wheat seeds. |
| 16:03 |
ShadowNinja |
rdococ: This is more suited for #minetest, but check the farming mod code or the wiki. |
| 16:03 |
ShadowNinja |
(probably farming:wheat_seeds) |
| 16:03 |
rdococ |
Well, it's for a mod, so it's development |
| 16:04 |
celeron55 |
https://github.com/minetest/minetest_game/blob/master/mods/farming/init.lua#L321 |
| 16:04 |
ShadowNinja |
rdococ: This channel is for engine development. |
| 16:04 |
celeron55 |
almost everything about minetest is development; that's why this channel is only about certain development 8) |
| 16:04 |
ShadowNinja |
(aka the core) |
| 16:05 |
rdococ |
Oh. Thanks for clarification. |
| 16:32 |
VanessaE |
speaking of certain types of developments... |
| 16:33 |
VanessaE |
has this been addressed yet? http://pastebin.ubuntu.com/7102951/ |
| 16:33 |
VanessaE |
seems to be new |
| 16:34 |
VanessaE |
but I'm a day or two behind on core |
| 16:35 |
troller |
old bug, generate sand float isand and touch it.. |
| 16:35 |
troller |
it was fixed in <censored> ;0 |
| 16:35 |
VanessaE |
troller: I'vegenerated sand blocks with tens of thousands of nodes without ever causing that. |
| 16:35 |
VanessaE |
(and made it fall of course) |
| 16:35 |
troller |
need mooore |
| 16:36 |
troller |
like 30x30x30 |
| 16:38 |
VanessaE |
troller: I'm being told there is no sand or gravel islands in the area causing it |
| 16:39 |
VanessaE |
but there are quarries from Technic operating nearby, could be digging into some underground gravel maybe |
| 16:39 |
VanessaE |
bu that's hardly "30x30x30" node.a |
| 16:39 |
VanessaE |
nodes* |
| 16:45 |
rdococ |
Why are the icons in the inventory all mixed up? How do I fix it? |
| 16:48 |
sfan5 |
are you using Windows? |
| 16:48 |
sfan5 |
(it's a common problem there) |
| 16:48 |
rdococ |
Yeah, and why does it bug in windows? |
| 16:48 |
rdococ |
I tried installing for ubuntu but it crashed |
| 16:48 |
rdococ |
I want to fix it as it's so annoying |
| 16:49 |
sfan5 |
if it crashes it would be useful to report that on the forums/github |
| 16:49 |
rdococ |
No what I mean is that my ubuntu crashes frequently and it just happened to while installing minetest |
| 16:49 |
sfan5 |
ah |
| 16:50 |
rdococ |
I'll try again someday |
| 16:50 |
hmmmm |
hmm |
| 16:50 |
hmmmm |
guys, we need a two-man system for backporting any bugfixes from freeminer |
| 16:51 |
hmmmm |
one person needs to find the commit that fixed it and describe at a high level what the problem was that it solved |
| 16:51 |
hmmmm |
and then another, separate person needs to actually implement a bugfix from that description of the problem alone |
| 16:52 |
hmmmm |
that second person should not be allowed to view freeminer's commits during the time they're implementing the fix |
| 16:53 |
hmmmm |
(of course, this is assuming freeminer has any fixes... usually freeminer's developments cause more bugs |
| 16:53 |
troller |
need more peoples for oneline fixes! |
| 16:53 |
troller |
hmmmm, what bugs of fm do you know? |
| 16:57 |
nore |
hmmmm, what was done here was to add a delay when nodeupdate makes recursive calls |
| 16:57 |
VanessaE |
ok guys, my logs shows this crash keeps happening when someone digs stuff like farming:cotton_8 |
| 16:57 |
VanessaE |
wtf? |
| 16:58 |
nore |
now that I've seen the commit, I'm not allowed to code the fix, right? |
| 16:58 |
VanessaE |
nore: it would be useful to know just how deep the recursion went before the crash too |
| 16:59 |
nore |
VanessaE, did you try to make a 1x1x100 line of sand fall? |
| 16:59 |
nore |
(perhaps x500) |
| 17:00 |
VanessaE |
nore: I did not. but in the past, i've done things like 50x1x50 without failures like this |
| 17:00 |
VanessaE |
and even 200x1x200 |
| 17:00 |
VanessaE |
THIS is happening during routine gameplay |
| 17:01 |
nore |
do you have a reliable way of reproducing the bug? |
| 17:01 |
VanessaE |
I used to generate and make-fall "hundreds x 1 x hundreds" sand or gravel islands as a hacky way to fix bad lighting |
| 17:02 |
VanessaE |
nore: that's just it, I haven't the foggiest idea what's doing it. |
| 17:02 |
VanessaE |
take a look at this: |
| 17:02 |
VanessaE |
(wait) |
| 17:03 |
rdococ |
Can we make a version of minetest for mobile devices, like Minecraft PE but never lags? |
| 17:03 |
hmmmm |
you say "we" like you're having a part in doing it |
| 17:04 |
hmmmm |
and yes, an *official* android port is almost complete, thanks to sapier |
| 17:04 |
VanessaE |
nore: http://pastebin.ubuntu.com/7103132/ |
| 17:04 |
hmmmm |
nore: thank you. you can't do it though, otherwise that might be construed as a copyright violation |
| 17:04 |
VanessaE |
hmmmm: sapier's build finally runs on my little piece of crap tablet :) |
| 17:06 |
hmmmm |
runs, but does it run well... |
| 17:06 |
VanessaE |
hmmmm: about as well as buildcraft did |
| 17:06 |
hmmmm |
i personally think it's impossible to get minetest to run well enough on these cpu-bottlenecked devices |
| 17:06 |
VanessaE |
I didn't get a lot of time to test it |
| 17:06 |
VanessaE |
BUT |
| 17:06 |
VanessaE |
it works better than BC |
| 17:06 |
hmmmm |
of course |
| 17:06 |
troller |
need to make looot of optimizations to run well |
| 17:07 |
hmmmm |
BC is just some hacked together crap to make a quick buck |
| 17:07 |
hmmmm |
did anybody look into DMCA takedowns for the samsung appstore and that other thing? |
| 17:07 |
hmmmm |
i think it was amazon |
| 17:07 |
hmmmm |
who here handles the legal issues? |
| 17:10 |
|
EvergreenTree joined #minetest-dev |
| 17:15 |
nore |
I made a 60^3 sand island fall ~10 mins ago... the server is still using all my CPU... |
| 17:15 |
VanessaE |
I'm being told now that even if there are technic quarries operating and digging into sand or gravel, they don't cause those nodes to fall so this must be a normal user dig action |
| 17:18 |
|
Calinou joined #minetest-dev |
| 17:18 |
nore |
ah, it works again, a lot of entities are being destroyed since they got above the limit of objects per block |
| 17:20 |
|
PenguinDad joined #minetest-dev |
| 17:20 |
celeron55 |
hmmmm: nobody anymore; doing it is worthless |
| 17:20 |
nore |
btw: what may perhaps cause a crash is that entity id is an u16... and you get above this limit when making so much sand fall |
| 17:20 |
celeron55 |
just get the official port going and only an insignificant number of people will look at the other crapware |
| 17:21 |
nore |
ok, now it looks like the server is working again, but I get a lot of 18:20:43: ERROR[ServerThread]: Server::ProcessData(): Cancelling: No player for peer_id=2 |
| 17:22 |
nore |
and mainloop still eats a lot of time: I have fps = 6 with drawtime = 30 |
| 17:23 |
nore |
(0.3 for mainloop_dtime and 0.3 for mainloop_other) |
| 17:25 |
hmmmm |
I think I got an idea |
| 17:25 |
hmmmm |
celeron: re our discussion from last night |
| 17:25 |
hmmmm |
have you ever seen a super-simplified gaussian blur approximation where it just calculates two convolutions of a flat blur kernel |
| 17:26 |
hmmmm |
what it does is keeps a running total of each value in the immediate radius in the X direction and sets the pixel at the curernt point to that running total / number of samples |
| 17:26 |
hmmmm |
that is the definition of "mean" |
| 17:27 |
VanessaE |
nore: seems that there needs to be water nearby |
| 17:27 |
hmmmm |
so we calculate this simplified gaussian of a chunk and set that aside |
| 17:27 |
|
Hiradur joined #minetest-dev |
| 17:27 |
nore |
VanessaE, for what? |
| 17:27 |
hmmmm |
now, statistical variance is defined as sum(p_i * (x_i - mean)^2 |
| 17:27 |
hmmmm |
we rearrange this to the simplified equation of |
| 17:28 |
hmmmm |
sum((p_i * x_i^2)) - mean^2 |
| 17:28 |
VanessaE |
nore: for the crash, all I'm getting so far from twoelk is water, dirt, and sand/graven seem to be needed in close proximity. He's now lost a water source that was in the ground, but the flowing water that it produced is still present. |
| 17:28 |
hmmmm |
our samples all have equal weights so we can simplify this to |
| 17:28 |
hmmmm |
sum(x_i^2) / N - mean^2 |
| 17:28 |
hmmmm |
remember the previous gaussian blur calculation? |
| 17:29 |
hmmmm |
square each value and you have that second component |
| 17:29 |
hmmmm |
now all you have to do is, when keeping the running total, square each value before adding it in or subtracting it out |
| 17:29 |
VanessaE |
gravel* |
| 17:30 |
hmmmm |
this comes to a total of 4 sweeps over each value in the chunk for a decent approximation of the variance blur that I wanted |
| 17:30 |
hmmmm |
err, I didn't highlight you... celeron55 |
| 17:30 |
celeron55 |
>when keeping the running total, square each value before adding it in or subtracting it out |
| 17:30 |
celeron55 |
does that actually work |
| 17:31 |
hmmmm |
you have your running total, call it h |
| 17:31 |
|
Shardvex joined #minetest-dev |
| 17:31 |
hmmmm |
h -= (map[i - radius / 2] * map[i - radius / 2]); |
| 17:31 |
hmmmm |
and now |
| 17:32 |
celeron55 |
but whatever, doing the squaring in a second step is totally feasible too so i don't need to care (ehm... i think) |
| 17:32 |
hmmmm |
h += (map[i + radius / 2] * map[i + radius / 2]); |
| 17:32 |
hmmmm |
output[i] = (h / radius); |
| 17:33 |
celeron55 |
so, given that that works, what are you going to do then |
| 17:33 |
celeron55 |
you have your variance |
| 17:33 |
hmmmm |
that's difficult because you have to overcalculate the map |
| 17:33 |
hmmmm |
this wouldn't be much of a problem if we didn't have 3d noise |
| 17:35 |
hmmmm |
if there's 3d noise, you can't just calculate a 2d height map, you need to poll every single point in the space which is much, much more computationally intensive |
| 17:36 |
celeron55 |
hmm, true |
| 17:36 |
nore |
now I get that: 18:35:07: ERROR[main]: ClientEnvironment::processActiveObjectMessage(): id=1 type=e SerializationError in processMessage(), message="" |
| 17:37 |
hmmmm |
ahhhh |
| 17:37 |
hmmmm |
but that's actually not a real problem |
| 17:38 |
hmmmm |
because i maintain a heightmap after all terrain modifications |
| 17:38 |
hmmmm |
after all phases, 2d noise, 3d positive, 3d negative, discrete features, and then cave carving, I do a plain top-down scan until I hit a solid block |
| 17:39 |
hmmmm |
this is in the mapgen phase |
| 17:39 |
hmmmm |
er nevermind, because this still necessitates overgenerating the entire cubic area |
| 17:40 |
hmmmm |
if we go ahead with overgeneration, it might be a good idea to solve the cave problem once and for all at the same time |
| 17:40 |
hmmmm |
but this is going to make map generation noticably slower |
| 17:40 |
celeron55 |
well i guess the question is, how much overgeneration is needed |
| 17:40 |
|
salamanderrake joined #minetest-dev |
| 17:40 |
hmmmm |
one block in all directions |
| 17:41 |
celeron55 |
is that enough? |
| 17:41 |
hmmmm |
of course |
| 17:41 |
celeron55 |
i don't think it's obvious at all |
| 17:41 |
hmmmm |
i don't plan on making terrain variance radiuses greater than 32 |
| 17:41 |
celeron55 |
a large-scale variation could be useful too; with a larger radius than 16 |
| 17:41 |
hmmmm |
no, 32 |
| 17:41 |
hmmmm |
because it's halved, remember? |
| 17:42 |
hmmmm |
only half of the radius sticks out |
| 17:42 |
celeron55 |
oh well whatever, sampling radius anyway |
| 17:42 |
hmmmm |
large-scale variation is a much, much more difficult problem |
| 17:42 |
hmmmm |
i don't want to go there |
| 17:42 |
celeron55 |
if not mapblocks, then the option would be to buffer generation as full chunks |
| 17:43 |
celeron55 |
but that's probably too unresponsive |
| 17:43 |
hmmmm |
at that point it's absolutely necessary to go with a perlin noise manipulation-based approach for getting variance |
| 17:43 |
hmmmm |
which dictates mapgen design |
| 17:43 |
celeron55 |
okay i think that option can be closed as infeasible then |
| 17:44 |
hmmmm |
you present an interesting idea |
| 17:44 |
hmmmm |
so what if terrain was done in a separate phase than the rest of it |
| 17:44 |
hmmmm |
whenever a chunk is requested for generation, it'll generate the terrain only of all the neighbors |
| 17:44 |
celeron55 |
it would potentially be simpler than the block one |
| 17:44 |
hmmmm |
when it goes to actually request generation of those blocks |
| 17:45 |
hmmmm |
you just calculate the biomes and set the top nodes and add vegetation |
| 17:47 |
hmmmm |
so blocks can have three states |
| 17:47 |
hmmmm |
ungenerated, terrain formed, decorated |
| 17:47 |
hmmmm |
decorated is the final state |
| 17:48 |
celeron55 |
do we need to actually care whether only strips of blocks are left in the terrain state or whether it's full chunks |
| 17:48 |
hmmmm |
this is easy to talk about but it'd mark a rather large change in the way minetest works |
| 17:48 |
celeron55 |
maybe not |
| 17:48 |
hmmmm |
good point |
| 17:49 |
hmmmm |
we could get away with only doing this for mapblocks |
| 17:49 |
hmmmm |
then that 3x3x3 inner mapblock area of each chunk, they would be completely generated in one single phase |
| 17:49 |
celeron55 |
well it's going to work in one of those ways and nothing else, but it may not matter which is chosen in the end |
| 17:50 |
VanessaE |
nore: [03-16 13:49] <twoelk> the server seems to crash if I dig in a certain area, or water touches those nodes |
| 17:50 |
hmmmm |
cool |
| 17:50 |
hmmmm |
it works in concept |
| 17:50 |
hmmmm |
now how do we parallelize it |
| 17:51 |
celeron55 |
there are still the old issues of merging decorations and lighting at the edges |
| 17:51 |
hmmmm |
LIGHTING. |
| 17:52 |
hmmmm |
ugh lighting |
| 17:52 |
hmmmm |
kill me now |
| 17:52 |
troller |
remove lighting from server! |
| 17:52 |
celeron55 |
if that happens, then this is the 0.5 branch |
| 17:52 |
hmmmm |
realbadangel is working on hardware lighting |
| 17:53 |
hmmmm |
that sort of development, combined with removing param1, would be 0.5.0 worthy |
| 17:53 |
hmmmm |
repurposing param1 rather |
| 17:53 |
celeron55 |
i don't think in terms of "worthy", i think in terms of "necessary" |
| 17:54 |
celeron55 |
(such a thing is going to take time to mature) |
| 17:54 |
celeron55 |
(while being incompatible) |
| 17:55 |
celeron55 |
(while being used) |
| 17:55 |
hmmmm |
good point |
| 17:56 |
hmmmm |
hmm |
| 17:56 |
hmmmm |
well I know what I need to work on |
| 17:57 |
hmmmm |
thing is, where am I going to find the time to do this :D I'm in the middle of switching jobs and all this nonsense |
| 17:57 |
celeron55 |
ask less hours in the new job |
| 17:58 |
hmmmm |
lol |
| 17:58 |
celeron55 |
some people say though that working for free is stupid; it's what tends to happen when working on minetest |
| 17:58 |
hmmmm |
i don't know if I can work part time at other companies |
| 17:58 |
celeron55 |
i have a four-day work week at the moment; it's nice |
| 17:59 |
hmmmm |
here, I can, but that's rare |
| 17:59 |
hmmmm |
they also allow overtime which is extremely rare (but you need to be authorized to use it) |
| 17:59 |
celeron55 |
if i could, i would pay for you, but there's no money moving anywhere regarding to minetest so that won't happen |
| 18:00 |
hmmmm |
d'aw thanks <3 |
| 18:00 |
celeron55 |
(in case that's not obvious to somebody...) |
| 18:00 |
celeron55 |
i would pay myself also, but that won't happen either |
| 18:02 |
hmmmm |
you know I was thinking, what if we switch back over to GPL and then use a dual licensing scheme so commercial users would have to pay for licenses |
| 18:02 |
hmmmm |
if Minetest gets even more generic than it is right now, and really good, it might be a viable game engine |
| 18:03 |
hmmmm |
like unity except for block-based-bullshit |
| 18:04 |
celeron55 |
if minetest licenses would be sold, then there needs to be a for-profit or not-for-profit organization doing that |
| 18:04 |
|
restcoser joined #minetest-dev |
| 18:04 |
celeron55 |
so that the legal stuff can be actually sorted out |
| 18:05 |
celeron55 |
but there's an issue in that scheme: the sold thing would only be under LGPLv2 |
| 18:05 |
celeron55 |
not more free |
| 18:06 |
celeron55 |
unless a lot of legally very questionable decisions were made |
| 18:07 |
hmmmm |
hmm |
| 18:07 |
hmmmm |
true, because then you'd need to discuss this with everybody who has made contributions and have them agree or tear their code out |
| 18:07 |
Calinou |
<hmmmm> you know I was thinking, what if we switch back over to GPL and then use a dual licensing scheme so commercial users would have to pay for licenses |
| 18:07 |
celeron55 |
of course if many people are interested in using a lot of time for such (there *is* some money to be made there), then a completely rewrite is totally possible |
| 18:07 |
Calinou |
commercial usage is already allowed; don't mix commercial an proprietary |
| 18:07 |
celeron55 |
but only if people work like in a whatever startup |
| 18:07 |
* Calinou |
builds a MG42 in the channel... |
| 18:07 |
hmmmm |
right |
| 18:08 |
hmmmm |
this would be a startup kind of thing |
| 18:08 |
hmmmm |
at this point we'd need to look around for VC funding |
| 18:08 |
Calinou |
also, is it really useful to be a game engine just about voxels? there doesn't seem to be much kind of games you can make out of it |
| 18:08 |
|
indriany joined #minetest-dev |
| 18:08 |
hmmmm |
calinou, voxels are just what you see |
| 18:09 |
hmmmm |
there's nothing stopping you from making sloped nodes and whatever |
| 18:09 |
celeron55 |
funding isn't probably *that* hard; for example the EU is currently giving money to new game projects (something like 50% them, 50% bank) |
| 18:09 |
|
twoelk joined #minetest-dev |
| 18:09 |
hmmmm |
celeron, funding is really difficult here in 'murrika if you don't have a solid business plan and angel investors |
| 18:09 |
Calinou |
crowdfunding is tricky around voxel-based games right now |
| 18:10 |
hmmmm |
how would that even work? |
| 18:10 |
Calinou |
lots of scams ;) |
| 18:10 |
hmmmm |
not crowdfunding |
| 18:10 |
celeron55 |
well minecraft-like games are going out of fashion at the moment |
| 18:10 |
celeron55 |
so obviously getting anyone interested is very hard |
| 18:10 |
hmmmm |
as a whole, the gaming industry is low profit margin |
| 18:11 |
hmmmm |
it makes me feel like the whole thing would be a gigantic waste of effort for just breaking even |
| 18:11 |
celeron55 |
yes unless you're making scammy mobile game |
| 18:11 |
celeron55 |
+ |
| 18:11 |
celeron55 |
s |
| 18:11 |
hmmmm |
hah |
| 18:11 |
hmmmm |
oh so true |
| 18:11 |
hmmmm |
we'd have to rename minetest to zyngtest or something and then load it with ads and do data mining on individuals' private information |
| 18:12 |
celeron55 |
mese pickaxe for $1 |
| 18:12 |
celeron55 |
instant monies |
| 18:12 |
hmmmm |
awesome |
| 18:12 |
hmmmm |
microtransactions are perfect for minetest actually |
| 18:12 |
hmmmm |
just that we'd need an "official" network |
| 18:13 |
hmmmm |
none of this works well with free software though |
| 18:13 |
|
iqualfragile joined #minetest-dev |
| 18:13 |
celeron55 |
if we want to get rich, we could set up "free" servers where people can buy items |
| 18:13 |
celeron55 |
it would actually work, right away |
| 18:13 |
hmmmm |
well, by 'rich' you mean like $50 tops |
| 18:14 |
hmmmm |
minetest's userbase is tinnnnny |
| 18:14 |
celeron55 |
well yeah, it would require a lot of marketing |
| 18:14 |
hmmmm |
running a business is hard work |
| 18:14 |
hmmmm |
and then we'd have to decide how contributors fit into all of this |
| 18:14 |
hmmmm |
would we turn into something like a redhat except for FOSS games? |
| 18:15 |
twoelk |
nore: regarding crashes on VE-Creative. Digging in other places is possible, protected or not. The server crash seems to be restricted to digging in a certain region. Waterflow seems to not be vital for the crash but might favor it. |
| 18:15 |
celeron55 |
i'm not going to do that though; i don't have a need to scam people for the sake of it |
| 18:15 |
celeron55 |
or for the sake if getting money |
| 18:15 |
|
rubenwardy joined #minetest-dev |
| 18:15 |
hmmmm |
i'll agree that there are simply too many problems with this idea |
| 18:15 |
hmmmm |
but keep up the entrepenurial spirit. passive income is a necessity these days |
| 18:17 |
ShadowNinja |
<ShadowNinja> Where was that thing that we set up where you could make a donation for a particular issue? It didn't get mentioned anywhere but this channel so it was never used, but maybe it would if we mentioned it on the donations page. |
| 18:17 |
sapier |
https://github.com/sapier/minetest/tree/android_2 |
| 18:18 |
hmmmm |
there's no way anybody would donate more than 10 cents to minetest |
| 18:18 |
hmmmm |
it's truly a labor of love |
| 18:18 |
sapier |
http://animalsmod.comuf.com/downloads/Minetest-debug.apk |
| 18:18 |
troller |
now MT have too low quality and alpha status to earn money |
| 18:18 |
celeron55 |
ShadowNinja: bountysource |
| 18:19 |
celeron55 |
hmmmm: well, when i have the donation page open on the website, people donate enough to cover the hosting and a bit more (i don't host the forum though) |
| 18:20 |
celeron55 |
which is, around $100 a year, total |
| 18:20 |
sapier |
could someone please look at the ngmm* files, I've done a major refactoring of mainmenu code. Imho the new variant is much more clean and consistent. But as always it may be personal preference so your opinion is welcome |
| 18:20 |
celeron55 |
currently it's closed because it seriously collects more money than i have any good use for |
| 18:21 |
hmmmm |
fair enough |
| 18:21 |
iqualfragile |
the android version could be sould |
| 18:21 |
iqualfragile |
would even be legal |
| 18:21 |
sapier |
basicaly new code adds tabviews and dialogs which can be visible or not. A tabview is standalone while a dialog always has to have a parent to return to |
| 18:21 |
hmmmm |
nobody's trying to make a profit on minetest, we decided that a while ago :) |
| 18:21 |
sapier |
tabviews contain tabs .. surprise surprise ;-) |
| 18:21 |
celeron55 |
so there's probably potential in getting a bit of money; but definitely not anywhere close to a wage or something; more like an occasional cup of coffee |
| 18:22 |
celeron55 |
for everyone doing coding at a given moment |
| 18:22 |
hmmmm |
wish i had an android device to try sapier's port on |
| 18:22 |
* iqualfragile |
too |
| 18:22 |
celeron55 |
of course we could pour the money to people who live in poor countries where it makes more difference |
| 18:22 |
sapier |
maybe some sort of developer meeting? ... well that's gonna be difficult I know |
| 18:22 |
hmmmm |
heh |
| 18:22 |
VanessaE |
sapier: dream on :P |
| 18:23 |
rubenwardy |
Testcon |
| 18:23 |
rubenwardy |
We all live to far apart to do that :P |
| 18:23 |
sapier |
I like celerons idea |
| 18:23 |
celeron55 |
we had one! there was me and iqualfragile! |
| 18:23 |
iqualfragile |
while i am no developer i actually went to finnland, its not that far away |
| 18:23 |
iqualfragile |
yep |
| 18:23 |
hmmmm |
and here I am having a hard enough time trying to convince my company to send me to defcon :) |
| 18:24 |
sapier |
finnland ... I'm not sure if I wanna be that close to russia the next months ;-) |
| 18:24 |
troller |
here too many bears! |
| 18:24 |
sapier |
btw does anyone have an idea why modstore is broken atm? |
| 18:25 |
celeron55 |
one option: we could set a price for the android version in the app store, but provide it for free on our website |
| 18:26 |
celeron55 |
then lazy rich people would pay for it and others could get it for no money |
| 18:26 |
sapier |
I assume appstore doesn't support "pay as much as you want" ? |
| 18:26 |
hmmmm |
awesome |
| 18:26 |
hmmmm |
let's do it |
| 18:26 |
celeron55 |
then we need some way to handle the money; am i the one who does that? |
| 18:27 |
hmmmm |
sapier should get half the profits :p |
| 18:27 |
sapier |
who who slow down guys, it's not yet complete ... there's some polishing to do |
| 18:27 |
celeron55 |
at least i've been doing all of the money stuff until this point and apparently people don't hate me yet for stealing it |
| 18:27 |
hmmmm |
still though, the android port took a LOT of work and it wouldn't ever exist without you. |
| 18:27 |
sapier |
no big things but takes time and doesn't show a lot of progress ... as always ;.) |
| 18:28 |
sapier |
there have been a lot of others contributing too |
| 18:28 |
twoelk |
mc took money for bugg-alpha/beta |
| 18:29 |
iqualfragile |
sapier: i gues the finns can defend themselves |
| 18:29 |
sfan5 |
>then lazy rich people would pay for it and others could get it for no money |
| 18:29 |
sfan5 |
s/lazy rich people/people who have been told that unknown app sources are really bad/ |
| 18:30 |
celeron55 |
it would go somehow like this: i pay for getting an account on google play, then after the cost of that is covered, sapier gets a good bunch of coffee money, and after that I need some planned way of using it (there may be anything between 0 to "too much") |
| 18:30 |
sfan5 |
hey, I want money too :-( |
| 18:30 |
iqualfragile |
everybody wants money |
| 18:30 |
iqualfragile |
sounds like a good idea |
| 18:30 |
hmmmm |
personally I'm ok |
| 18:30 |
celeron55 |
of course everyone wants money; that's the issue here |
| 18:30 |
sapier |
I'm for donating "too much" money |
| 18:31 |
iqualfragile |
or using it to host all of minetests presence first and then donating it |
| 18:31 |
hmmmm |
i hardly work on minetest these days |
| 18:31 |
sapier |
hmm guess I'm not "everyone" ;-) |
| 18:31 |
iqualfragile |
another option would be to use this bittip thingy |
| 18:31 |
sapier |
of course expenses need to be paid first |
| 18:31 |
hmmmm |
bittip or gittip? |
| 18:31 |
sfan5 |
how about both? |
| 18:32 |
celeron55 |
it's not really good because it encourages developers to try to look good instead of be useful |
| 18:32 |
hmmmm |
lol, that ashe dreyden gets $600/week from gittip and she does nothing but bitch on twitter about the evil white males |
| 18:32 |
celeron55 |
and again, nobody has actual use for the money that ensues from donations |
| 18:32 |
sapier |
the one who's got the expenses should be aware that once there's real money he's required to provide the docs about it ... yes that's quite normal for some people ... for others not |
| 18:33 |
iqualfragile |
yet another option would be to use it to reward the implementation of some more complex features |
| 18:34 |
hmmmm |
not sure how it works in the EU, but here you need to set up minetest as an NPO to make donations non-taxable |
| 18:34 |
celeron55 |
i roughly know the finnish law about this; if i'm doing it as a person, it's roughly legal if i pay income tax of it and then give it away as "gifts" |
| 18:34 |
hmmmm |
so there'd be quite a bit of legal legwork first |
| 18:34 |
hmmmm |
ew |
| 18:34 |
celeron55 |
(that's about the paid app) |
| 18:34 |
hmmmm |
you would not want to pay income tax on this |
| 18:34 |
sapier |
at least in germany it's same too |
| 18:34 |
hmmmm |
oh yeah |
| 18:34 |
VanessaE |
it also depends on just how much you get, too |
| 18:34 |
|
Megaf_ joined #minetest-dev |
| 18:34 |
hmmmm |
well if you set it up as a not-for-profit organization it'd be tax free as well |
| 18:34 |
sapier |
and there's plenty of paperwork to do each year to stay tax free |
| 18:35 |
VanessaE |
if this were US-based, if it's less than like $8000 a year or so, you don't have to report it at all |
| 18:35 |
celeron55 |
for donations the finnish law is rather vague, but my interpretation of it is that i don't need to do anything special for these donations that i have been handling now |
| 18:35 |
VanessaE |
(I forget the exact threshold) |
| 18:35 |
hmmmm |
the point is that there's no way in hell we should pay tax for minetest donations because our respective governments aren't doing jack shit and profiting off of something that we don't get compensated for anyway |
| 18:35 |
hmmmm |
and then what does it get spent on? |
| 18:35 |
hmmmm |
funding so we can get spied on illegally |
| 18:36 |
sapier |
but in germany we've got a saying "first hunt the animal prior donating it" ;) |
| 18:36 |
hmmmm |
so as a matter of efficiency and principle, we need to make sure any profit from minetest is not taxed |
| 18:36 |
celeron55 |
if a non-profit is formed, that should probably happen in germany because most people are from there |
| 18:36 |
iqualfragile |
hmmmm: or make it 'murica based because they have allmost no taxes there |
| 18:37 |
hmmmm |
hmm yes |
| 18:37 |
celeron55 |
(and germany is a roughly sane country) |
| 18:37 |
hmmmm |
the appstore purchases are technically donations |
| 18:37 |
hmmmm |
and if not, we change status to 'not for profit' |
| 18:37 |
sapier |
There are some minor things left: settings, npot2 images, https certificates, controls(decide about how to do it in final version) and a lot of testing ;-) |
| 18:37 |
hmmmm |
which is slightly different |
| 18:37 |
hmmmm |
iqualfragile: that's BS. there is a shitton of taxes here |
| 18:38 |
sapier |
and wchar handling, the current variant is quite hacky ;-) |
| 18:38 |
hmmmm |
if you fart or cough, there's a tax for that |
| 18:38 |
iqualfragile |
hmmmm: hahaha |
| 18:38 |
hmmmm |
and hardly any of it goes to anytihng useful |
| 18:38 |
sfan5 |
sapier: simply resize them with nearest-neighbor in irrlicht before loading them, I didn't see a quality-decreasement when doing that |
| 18:38 |
iqualfragile |
can you aproximate how much of your income gois into taxes? |
| 18:38 |
sfan5 |
>taxes >NSA |
| 18:39 |
hmmmm |
*my* income? not much, but that's because I don't make much to begin with and I'm getting a lot of deductions and credits this year because I was a student in 2013 and all this stuff |
| 18:39 |
sapier |
sfan5 I can't resize images in irrlicht because irrlicht can only resize textures ... and textures have to be npot2 on some architecures ;-) |
| 18:39 |
sapier |
or did I miss another way of resizing? |
| 18:40 |
sfan5 |
sapier: 1) implement nearest-neighbor for images in Irrlicht 2) ??? 3) Profit |
| 18:40 |
iqualfragile |
hmmmm: in germany its about 50% unless you realy do not get a lot |
| 18:40 |
iqualfragile |
in finnland its even more |
| 18:40 |
hmmmm |
gotta wonder how anybody is able to save |
| 18:40 |
celeron55 |
one option would be to beg to be part of sfconservancy.org or another similar entity, and handle donations through that |
| 18:40 |
iqualfragile |
so compared to that there are allmost no taxes in 'murica |
| 18:40 |
hmmmm |
true. |
| 18:41 |
hmmmm |
but those countries actually do things for you |
| 18:41 |
hmmmm |
murrica doesn't do shit |
| 18:41 |
iqualfragile |
i know |
| 18:41 |
hmmmm |
they just shit on your head and tell you to be grateful and be proud of your country |
| 18:41 |
iqualfragile |
well, it does run some wars, doesnt it? |
| 18:41 |
iqualfragile |
but this is getting ot |
| 18:41 |
celeron55 |
that all depends on how you calculate the total tax, which is practically impossible in a comparable and universal way |
| 18:41 |
VanessaE |
wars? :-/ |
| 18:41 |
sapier |
sfan5? sorry lost context for your last answer |
| 18:41 |
celeron55 |
so let's not talk about that |
| 18:42 |
hmmmm |
true |
| 18:42 |
celeron55 |
so any thoughts about sfconservancy.org? |
| 18:42 |
sfan5 |
sapier: add code to irrlicht to resize images |
| 18:42 |
hmmmm |
can't comment, i don't really know anything about it |
| 18:42 |
celeron55 |
the projects in there tend to be rather high tier |
| 18:42 |
hmmmm |
sfan5: read up, he mentioned why he can't |
| 18:43 |
celeron55 |
hmmmm: it's one of those non-profits that have been created in order to legally handle donations and legal issues for open source projects |
| 18:43 |
hmmmm |
ahh |
| 18:43 |
sapier |
I always can add code to resize images but that's the last thing to do |
| 18:43 |
hmmmm |
bretty gud |
| 18:43 |
hmmmm |
we might have to check it out |
| 18:43 |
sapier |
as I'd have to implement the image format myself |
| 18:43 |
sfan5 |
hmmmm: images, not textures; You should be able to resize images independently of what your GPU wants |
| 18:43 |
celeron55 |
the projects on sfconservancy.org are rather high tier (git, wine, whatever) |
| 18:43 |
celeron55 |
but we might find one that accepts minetest |
| 18:44 |
sfan5 |
sapier: Irrlicht gives you an interface to access the pixels |
| 18:44 |
sapier |
sfan5 for what I read that isn't possible |
| 18:44 |
iqualfragile |
well, we could allways ask |
| 18:44 |
sfan5 |
sapier: so you say my GPU has an effect on normal calculations my CPU does? |
| 18:45 |
celeron55 |
i have no idea what the goal should be with that either though |
| 18:45 |
iqualfragile |
"Project leaders […] are encouraged to apply for membership." |
| 18:45 |
sapier |
that was meant as irrlicht doesn't support resizing by default, of course we can do pixel based resizing but that's more then ugly |
| 18:45 |
sfan5 |
iqualfragile: pssst... almost* and always* |
| 18:46 |
sapier |
we're not a image manipulation lib ;-) |
| 18:46 |
sfan5 |
why is pixel based resizing ugly? |
| 18:46 |
iqualfragile |
sfan5: i faill to see the probllem |
| 18:47 |
sapier |
because we have to do the subpixel calculation on our own or get really really ugly results |
| 18:47 |
celeron55 |
sapier: you can get an image from irrlicht which you can resize as a bitmap and then you can insert it into irrlicht and get a texture |
| 18:47 |
celeron55 |
that's what minetest's texture modifiers do |
| 18:47 |
|
jin_xi joined #minetest-dev |
| 18:47 |
celeron55 |
and yes, you need to use a higher-resolution texture for upscaled things |
| 18:48 |
iqualfragile |
look like minetest kinda meets all the criteria, vibrant comunity being the only "critcal" point |
| 18:48 |
celeron55 |
an alternative is to just blit the image to a 2^n image and then use different texture coordinates, but that's quite complicated |
| 18:49 |
sapier |
I'm for resizing official minetest logo to 128x128 ;-) |
| 18:49 |
celeron55 |
(pixel-to-pixel, leaving blank space, in that case) |
| 18:50 |
sapier |
hmm I was hoping for some "use function xyz"-like answer ;-) |
| 18:50 |
iqualfragile |
sapier: if that realy is the only thing you would need it for: why not prescale it? |
| 18:50 |
|
Megaf joined #minetest-dev |
| 18:51 |
sapier |
I haven't tried yet but I assume mesh textures and mod screenshots to be the usecases where non npot2 textures occur |
| 18:51 |
ShadowNinja |
sapier: The icon's SVG. |
| 18:51 |
celeron55 |
sapier: i have heard that GPUs work better for textures that have power-of-two dimensions, even if they aren't the same (eg. 32x64); is this the case for android devices? |
| 18:52 |
sfan5 |
celeron55: not for all devices; only for some |
| 18:52 |
sapier |
possible but some gpu's just fail completely for non npot2 textures |
| 18:52 |
sapier |
they just don't load them |
| 18:52 |
celeron55 |
sapier: for gui images performance isn't an issue, just rescale the pixels |
| 18:52 |
sfan5 |
iOS (atleast the GPU in my iTouch 5) wants npot2 textures too |
| 18:52 |
celeron55 |
just add some special handling to tile.cpp |
| 18:53 |
celeron55 |
or alternatively add a power-of-two-izer texture modifier that is specially applied everywhere on android |
| 18:53 |
sapier |
well I know where to add as I already added a warning there :-) |
| 18:53 |
celeron55 |
just make sure in a way or another that TextureSource doesn't hand out non-power-of-two textures on android; problem solved |
| 18:54 |
celeron55 |
most performance-critical textures already are 2^n so that's not a problem |
| 18:55 |
sfan5 |
sapier: does sound actually WORK on android? |
| 18:55 |
sfan5 |
openal does not seem to find any real backends for me |
| 18:55 |
sfan5 |
only WaveFile and Null |
| 18:56 |
sapier |
yes |
| 18:56 |
sapier |
at least I hear my steps on grass |
| 18:56 |
sfan5 |
hm |
| 18:56 |
sfan5 |
did you use official openal srcs |
| 18:56 |
sfan5 |
? |
| 18:56 |
sfan5 |
hm |
| 18:57 |
sfan5 |
seems like -DLIBTYPE=STATIC makes OpenAL be a static library |
| 18:57 |
sfan5 |
Linking C static library libopenal.a |
| 18:57 |
sfan5 |
\o/ |
| 18:58 |
sapier |
I did use the libs that work and I couldn't get some libs to build static ... well if I did spend some more days maybe but it wasn't worth the work for me |
| 18:58 |
sapier |
otheres built static but not their dependencys e.g. curl and ssl |
| 18:59 |
sfan5 |
curl does not build static? |
| 19:00 |
celeron55 |
iqualfragile: the issue is, they probably want to know where the project is going |
| 19:00 |
sapier |
curl does but openssl doesn't at least I didn't manage to make curl use the static linked openssl |
| 19:00 |
iqualfragile |
celeron55: yeah, just think of some stuff |
| 19:00 |
iqualfragile |
(xyz|imag1nary please cut this part from the logs :D) |
| 19:01 |
sfan5 |
sapier: isn't openssl already included with android? |
| 19:01 |
sfan5 |
do we even need ssl support? |
| 19:01 |
sapier |
there are about 10 openssl-android trees out there exactly one very outdated one uses static linking |
| 19:01 |
iqualfragile |
android release would be a big point, wouldnt it? |
| 19:01 |
sapier |
well someone decided to switch mmdb to https some months ago ... so we need it |
| 19:01 |
sfan5 |
oh |
| 19:01 |
sfan5 |
mh |
| 19:02 |
sapier |
didn't realize this till yesterday too |
| 19:02 |
sfan5 |
that explains why that does not work |
| 19:02 |
celeron55 |
well, i can always reference the blog post that i made |
| 19:02 |
iqualfragile |
sfan5: no it does not |
| 19:02 |
celeron55 |
but there's no roadmap or such |
| 19:02 |
sfan5 |
no? |
| 19:02 |
iqualfragile |
the https change happened months ago |
| 19:02 |
sfan5 |
iqualfragile: I mean on android |
| 19:02 |
sapier |
no need to defend it iqualfragile it's fine the way it is it's just some additional work :) |
| 19:02 |
iqualfragile |
oh, right, thats true |
| 19:03 |
sapier |
I didn't find a way to use androids openssl they don't provide access for ndk |
| 19:03 |
sfan5 |
well |
| 19:03 |
sfan5 |
gnutls should be easier to cross-compile |
| 19:04 |
iqualfragile |
take it AWAAAY |
| 19:04 |
celeron55 |
remember to use a new enough gnutls 8) |
| 19:04 |
celeron55 |
they had at least that one nice bug |
| 19:04 |
sapier |
easier then entering ndk-build? |
| 19:04 |
iqualfragile |
gnutls seems to love aquiring bugs |
| 19:04 |
iqualfragile |
the bad ones |
| 19:04 |
sapier |
the issue is not building the lib but telling curl build to use it correctly |
| 19:04 |
sapier |
for some reason this is way more easy for shared libs |
| 19:05 |
sapier |
maybe because shared libs are the 99% case on linux like os |
| 19:06 |
sfan5 |
well |
| 19:06 |
sfan5 |
what will we do on iOS then |
| 19:06 |
sfan5 |
no shared libs there |
| 19:07 |
sapier |
I'm not gonna write a ios port ... but I consider transition to static libs to be polishing |
| 19:07 |
iqualfragile |
sfan5: hmm… is the ios hardware even powerfull enought? |
| 19:07 |
sfan5 |
iqualfragile: yes, it should be |
| 19:07 |
sapier |
the hardware will be but you may not get enough access to it |
| 19:07 |
sfan5 |
Apple build pretty powerful hardware into their devices |
| 19:07 |
sfan5 |
builds* |
| 19:11 |
sapier |
silly question but how to calculate pot2 for a arbitrary number e.g. 3445? |
| 19:11 |
sapier |
despite of consecutive dividing by 2 |
| 19:11 |
sfan5 |
why would you need that |
| 19:11 |
sfan5 |
? |
| 19:12 |
sapier |
to get the size of the image to scale it to? |
| 19:12 |
sfan5 |
simply resize the texture to width + (width % 2); height + (height % 2) |
| 19:12 |
sfan5 |
oh wait |
| 19:12 |
sapier |
not modulo pot |
| 19:13 |
sapier |
53x19 --> 64x32 |
| 19:13 |
celeron55 |
determine the highest bit in the variable, then check if any other bits are set; if not, then it's 2^n, if so, then zero it and all other bits and set the bit that is one higher |
| 19:14 |
celeron55 |
http://stackoverflow.com/questions/1322510/given-an-integer-how-do-i-find-the-next-largest-power-of-two-using-bit-twiddlin |
| 19:14 |
celeron55 |
there are some practical solutions 8) |
| 19:15 |
celeron55 |
of course just iterating through the 2^ns is probably plenty fast enough though |
| 19:16 |
celeron55 |
there are only as many of those as there are bits |
| 19:16 |
celeron55 |
the topmost answer is clearly the best though |
| 19:17 |
sfan5 |
sapier: did you OpenAL build include support for OpenSL? |
| 19:20 |
sfan5 |
ok |
| 19:20 |
sfan5 |
found openal version that supports android |
| 19:20 |
sfan5 |
https://github.com/AerialX/openal-soft-android |
| 19:29 |
sapier |
sfan why don't you just look at my branch build/android/makefile ... it's all there |
| 19:31 |
sfan5 |
oh, rigth |
| 19:31 |
sfan5 |
right* |
| 19:35 |
sapier |
did anyone succeed in ogles2 and meshes yet? |
| 19:35 |
celeron55 |
in what? |
| 19:36 |
sapier |
there seems to be some mesh operation crashing if I switch to ogles2 |
| 19:36 |
celeron55 |
on sailfish, ogles2 works a bit |
| 19:36 |
celeron55 |
it doesn't crash, but half of the meshes aren't drawn |
| 19:36 |
sapier |
something in inventory because once I move all things from panel to other inventory in ogles1 mode I can enter the game |
| 19:37 |
celeron55 |
(sailfish-native i mean) |
| 19:37 |
sapier |
mabe same thing but more severe result |
| 19:37 |
celeron55 |
to me it seemed like irrlicht's gles2 support is just broken |
| 19:37 |
celeron55 |
like, not fully developed even |
| 19:38 |
sapier |
possible |
| 19:38 |
celeron55 |
should ask from them |
| 19:38 |
sapier |
I see some rendering glitches too e.g. white pixels blinking between nodes |
| 19:39 |
celeron55 |
that's probably a side-effect of the shader-implementation of the fixed pipeline |
| 19:39 |
celeron55 |
the texture sampling code in there takes white pixels outside of textures instead of wrapping or ingoring them or something |
| 19:39 |
celeron55 |
i'd guess |
| 19:39 |
celeron55 |
(of course that's just one possibility) |
| 19:40 |
sapier |
maybe I don't know very much about graphics programming :-) you could tell me almost anything ;-) |
| 19:40 |
sapier |
but I had to copy shaders to assets in order to get ogles2 work so this sounds reasonable |
| 19:41 |
celeron55 |
ogles2 can't render anything without having a shader do it, so irrlicht provides the default ones that do roughly the same thing as the fixed pipeline does (i.e. regular opengl without shaders) |
| 19:42 |
sapier |
ahh that does make sense |
| 19:45 |
|
salamanderrake joined #minetest-dev |
| 20:24 |
sapier |
argh I hate irrlicht |
| 20:24 |
sapier |
they lie about npot 2 support |
| 20:27 |
|
PenguinDad joined #minetest-dev |
| 21:01 |
|
NakedFury joined #minetest-dev |
| 21:20 |
|
EvergreenTree joined #minetest-dev |
| 21:21 |
sapier |
ok added non npot2 support |
| 21:21 |
sapier |
had to ask ogles myself because irrlicht claims limited apple npot2 support to be full npot2 support |
| 21:51 |
|
OldCoder joined #minetest-dev |
| 22:14 |
|
OldCoder joined #minetest-dev |
| 22:18 |
RealBadAngel |
hi all |
| 22:18 |
RealBadAngel |
so can we merge #1117 ? |
| 22:18 |
ShadowBot |
https://github.com/minetest/minetest/issues/1117 -- Normal maps generation on the fly. by RealBadAngel |
| 22:21 |
ShadowNinja |
Hmmm, I should test that. I can't for at least a few hours though. |
| 22:23 |
RealBadAngel |
anybody else wanna test it? |
| 22:26 |
sapier |
<< testing right now ... anything I have to look ? |
| 22:29 |
sapier |
do I need special textures? |
| 22:31 |
ShadowNinja |
sapier: Just make sure everything compiles and runs on your GPU. |
| 22:31 |
RealBadAngel |
just turn on shaders, bumpmapping and normal map generation |
| 22:32 |
sapier |
hmm I don't see leaves waving |
| 22:32 |
RealBadAngel |
2nd feature is overriding normals, https://forum.minetest.net/viewtopic.php?id=8714 |
| 22:33 |
RealBadAngel |
sapier, are you trying it with minetest_game ? |
| 22:44 |
|
PenguinDad joined #minetest-dev |
| 22:44 |
|
jin_xi joined #minetest-dev |
| 22:44 |
|
sapier joined #minetest-dev |
| 22:44 |
|
darkrose joined #minetest-dev |
| 22:45 |
sapier |
I hate net splits |
| 22:45 |
RealBadAngel |
shit happens |
| 22:46 |
RealBadAngel |
about leaves and plants waving. you need minetest_game for it, minimal wont do |
| 22:46 |
troller |
sapier, fiiix! |
| 22:46 |
troller |
02:44:58: ACTION[ServerThread]: Guest7863 [::ffff:23.117.228.43] joins game. |
| 22:46 |
troller |
02:44:58: ACTION[ServerThread]: Guest7863 [::ffff:23.117.228.43] joins game. List of players: |
| 22:46 |
PenguinDad |
I don't seem to be able to reproduce #1175 |
| 22:46 |
ShadowBot |
https://github.com/minetest/minetest/issues/1175 -- broken icon graphics with "preload item visuals" |
| 22:47 |
sapier |
troller if you did spend half of the time you spent on complaining for writing a fix you'd have a golden edge solution by now |
| 22:54 |
ShadowNinja |
troller: You can make a patch too. It's most likely a one-line change. |
| 22:54 |
troller |
2 lines. |
| 22:56 |
CiaranG |
PenguinDad: pretty sure 1175 depends on video drivers or something - I get that same thing on maybe 4 out of 8 machines. I just turned it off. |
| 22:56 |
troller |
https://github.com/freeminer/freeminer/commit/0216127ffef1647241b5e93f370015d1dd2af594 |
| 22:56 |
|
17SAAW0I4 joined #minetest-dev |
| 22:56 |
|
Zeitgeist_ joined #minetest-dev |
| 22:57 |
sapier |
DO NOT use those commits ... they're license incompatible to minetest |
| 22:58 |
sapier |
sorry proller but you have to either do a pull request to minetest or explicitly relicense your fixes lgpl if you do add them to freeminer only |
| 22:58 |
troller |
change one space ;) |
| 22:59 |
sapier |
ok :-) guess there's no big deal ... yet we can't directly merge them |
| 23:00 |
sapier |
but I guess writing functional equivalent code for this will not be a big deal ;-) |
| 23:05 |
RealBadAngel |
sapier, how testing goes? |
| 23:05 |
sapier |
we:-) despite of my sea sicknes it seems to be fine |
| 23:07 |
|
PenguinDad joined #minetest-dev |
| 23:09 |
iqualfragile |
sapier: why is it license incompatible? |
| 23:10 |
|
VargaD joined #minetest-dev |
| 23:10 |
sapier |
because freeminer switched to gpl which is less free then lgpl |
| 23:10 |
sapier |
this way proller can use minetest code in freeminer but not the other way round |
| 23:11 |
iqualfragile |
sapier: the LICENSE file in the freeminer repos still states gpl |
| 23:12 |
sapier |
as I said GPL not LGPL |
| 23:12 |
sapier |
you can't transfer GPL code to LGPL |
| 23:12 |
iqualfragile |
*lgpl |
| 23:13 |
sapier |
proller did you revert that change? |
| 23:14 |
troller |
no |
| 23:15 |
sapier |
hmm then you've got double licensing right now ;-) |
| 23:16 |
troller |
in LICENSE.txt nothing about freeminer |
| 23:18 |
sapier |
I thought I'd only have to deal with license issues at work now it's getting same crap here :-( |
| 23:20 |
PilzAdam |
sapier, as far as I see FM is under LGPL |
| 23:21 |
sapier |
PilzAdam: not entirely |
| 23:21 |
PilzAdam |
at least thats what their LICENSE.txt says |
| 23:21 |
sapier |
COPYING as well as headers of some of xyz's files specifyl gpl |
| 23:22 |
RealBadAngel |
sapier, you can try generatin normals on any texture pack |
| 23:24 |
PilzAdam |
can they distribute LGPL'd code under GPL without asking the orginal author? |
| 23:24 |
sapier |
yes |
| 23:25 |
sapier |
but most of those files are written by xyz anyway |
| 23:26 |
|
OldCoder joined #minetest-dev |
| 23:27 |
sapier |
and more relevant even if they couldn't change the license of existing code their additions would still be gpl |
| 23:27 |
PilzAdam |
they can only relicense it if they modify it |
| 23:27 |
sapier |
I don't se any reason to merge our own code back ;-) |
| 23:27 |
sapier |
we'd need the fixes but those are gpl |
| 23:39 |
iqualfragile |
uhm… no they actually can not do that |
| 23:40 |
iqualfragile |
wait, rethinking |
| 23:40 |
sapier |
that's what lgpl is for ;-) |
| 23:42 |
iqualfragile |
http://www.gnu.org/licenses/license-list.html#LGPL |
| 23:43 |
iqualfragile |
(i did not quite understand why minetest switched to lgpl anyways) |
| 23:44 |
sapier |
because lgpl doesn't have the copyleft |
| 23:44 |
sapier |
basicaly proller did was lgpl was meant for ... he didn't agree with mainline development and forked |
| 23:45 |
iqualfragile |
"was" ^^ |
| 23:45 |
sapier |
ok actually it wasn't proller but xyz |
| 23:45 |
iqualfragile |
you can do that with gpl, too |
| 23:45 |
sapier |
no you can't use gpl code in a lgpl project as this will result in the lgpl code beeing gpl |
| 23:46 |
iqualfragile |
again: i do not understand why minetest switched to lgpl |
| 23:46 |
iqualfragile |
switching to agpl would make more sense in my opinion |
| 23:46 |
celeron55 |
it doesn't matter whether you understand it or not |
| 23:46 |
celeron55 |
minetest uses lgpl and that's it |
| 23:46 |
sapier |
possible but we're way beyond the point where switching licenses is possible |
| 23:46 |
sapier |
at least for non compatible licenses :-) |
| 23:47 |
iqualfragile |
celeron55: iirc the license was changed a while ago, i would like to understand the reasons for that |
| 23:47 |
celeron55 |
iqualfragile: a while? it was changed years ago |
| 23:48 |
sapier |
and it doesn't change the situation minetest is lgpl and we can't switch to a non compatible license |
| 23:49 |
iqualfragile |
sapier: it would be kinda possible |
| 23:49 |
sapier |
if you did ask anyone who contributed to minetest |
| 23:49 |
celeron55 |
what about we switch to a license that disallows development and usage? |
| 23:50 |
iqualfragile |
celeron55: great idea! |
| 23:50 |
|
Hiradur joined #minetest-dev |
| 23:50 |
celeron55 |
because people seem to like licenses that restrict things |
| 23:50 |
sapier |
I don't see a reason to switch license lgpl is not perfect but good enough |
| 23:50 |
celeron55 |
now please everyone who cannot stand LGPL, leave this channel; it is not a thing that is arguable at all |
| 23:51 |
celeron55 |
i consider one of my greates work regarding to minetest to be the ninja'ed license switch from GPL to LGPL |
| 23:51 |
celeron55 |
+t |
| 23:51 |
sapier |
btw iqualfragile did you specify mmdb license yet? ;-) |
| 23:52 |
iqualfragile |
uhm… no |
| 23:52 |
iqualfragile |
:D |
| 23:52 |
iqualfragile |
licenses are hard |
| 23:53 |
sapier |
I suggest to decide for one we could realize by some time our mmdb and serverlist are non-free code ;-P |
| 23:54 |
iqualfragile |
hmm… i might go for agpl, but it will be hard to actually write that down as others (mainly jeija) contributed some stuffs and he is unreachable |
| 23:54 |
celeron55 |
(and i consider one of my worst work to be the lockdown to GPL and due to that LGPL because i knew almost nothing about licenses initially) |
| 23:56 |
celeron55 |
i was trying to go for BSD for contributions, but then i let it slip because the initial contributors were some GPL fanatics |
| 23:56 |
celeron55 |
which was the worst idea ever |
| 23:57 |
iqualfragile |
well, maybe not, with gpl those craptastic buildcraft apps would be legal |
| 23:58 |
iqualfragile |
s/gpl/bsd |
| 23:58 |
iqualfragile |
additionally bsd brought us apple -.- |
| 23:58 |
sapier |
without those apps we'd not have a android port |
| 23:59 |
sapier |
it's quite simple if those crapy apps didn't ruin out reputation I'd not have started to link the pieces |
| 23:59 |
sapier |
at least not that early |
