Time  Nick            Message
05:38 RealBadAngel    hi all
05:38 VanessaE        hey RBA
05:38 RealBadAngel    whats wrong with forums?
05:39 VanessaE        seems okay to me?
05:39 RealBadAngel    my browser says certificates are not valid for the site
05:39 RealBadAngel    and connection is "not trusted"
05:39 RealBadAngel    and i cannot view forums at all
05:39 VanessaE        odd
05:40 VanessaE        maybe you need to validate a new certificate?
05:40 RealBadAngel    forum.minetest.net używa nieprawidłowego certyfikatu bezpieczeństwa.
05:40 RealBadAngel    Ten certyfikat jest prawidłowym certyfikatem tylko dla następujących nazw:
05:40 RealBadAngel    users.minetest.ru , minetest.ru
05:40 VanessaE        thexyz was doing some maintenance earlier
05:40 RealBadAngel    site uses wrong certificate, this certificate is valid only for following names:
05:41 RealBadAngel    this happened yesterday
05:41 VanessaE        odd.
05:41 VanessaE        it's working fine for me now
05:41 RealBadAngel    i just came back from work and its still the same
05:43 RealBadAngel    .ru means that thexyz has something to do with for sure
05:43 RealBadAngel    firefox didnt pulled that out of the hat
08:11 nore            any thought on this? https://github.com/minetest/minetest_game/pull/198
08:14 nore            thexyz, sfan5: anything againt that ^ it is very basic
12:40 kahrl           so I tried the client media fetch code with 2 local apaches and wiresharked it
12:40 kahrl           (the new code in httpfetch, I mean)
12:41 kahrl           turns out it actually uses keepalive pretty well, glad I don't have to add more code to that problem :)
12:43 kahrl           I think it's curl_multi that does that (I only use 8 curl handles but it keeps 10 connections)
12:45 PilzAdam        RealBadAngel, the certificate is valid; the problem is in your browser
15:59 Taoki[laptop]   RealBadAngel: Any news on parallax and color sky fog?
16:00 RealBadAngel    nothing newer than 3 weeks ago
16:00 RealBadAngel    without online documentation i was simply idle
16:00 Taoki[laptop]   ok ;( Hope they will make it
16:01 RealBadAngel    i got it workin but to some degree
16:01 Taoki[laptop]   Why so? Are there things about the parallax code that need documentation to explain?
16:01 Taoki[laptop]   Looked great in the last screenshots to me
16:01 RealBadAngel    irrlicht engine docs
16:02 RealBadAngel    i need a way to pass to each tile (shader instance) calculated tangent vector
16:03 Taoki[laptop]   I see..
16:03 RealBadAngel    either we move with meshes to built in version that provides tangent and binormal (which means rebuilding of lotsa our code)
16:03 Taoki[laptop]   Might help asking in #irrlicht here or the irrlicht forum on sourceforge IIRC.
16:03 Taoki[laptop]   I don't know what to ask exactly so it wouldn't help for me to post and discuss
16:04 RealBadAngel    or we will find a way to pass vec4 (float) to shader
16:04 RealBadAngel    but not global, an attribute
16:04 Taoki[laptop]   Won't work as an Uniform variable
16:04 Taoki[laptop]   ?
16:04 RealBadAngel    wont
16:05 Taoki[laptop]   What about passing each value of the vec4 as a float?
16:05 RealBadAngel    uniform is for each and every instance of a mesh
16:05 Taoki[laptop]   ah
16:05 RealBadAngel    we need such vector for each every different tile
16:05 RealBadAngel    with different coords in the world
16:05 Taoki[laptop]   I think there was an alternative to uniform, but I don't remember well. Some way to pass values to the shader separately
16:06 RealBadAngel    its called attribute
16:06 RealBadAngel    sadly irrlicht doesnt support that
16:06 Taoki[laptop]   Yeah, that
16:06 RealBadAngel    thats why im trying to pass a data texture to shader
16:07 RealBadAngel    that i can easily bound to the instance
16:07 Taoki[laptop]   You posted screenshots that work though. Why doesn't that way work until this can be fixed?
16:08 RealBadAngel    tangent was calculated inside shader
16:08 RealBadAngel    that has two cons
16:08 RealBadAngel    heavily GPU usage
16:08 RealBadAngel    and is glitchy
16:08 Taoki[laptop]   I see
16:08 RealBadAngel    because you cannot calculate tangent properly out of one point
16:08 Taoki[laptop]   Personally I'd vote for putting that upstream till the better way is found, unless it's very very broken.
16:08 RealBadAngel    you need triangle to do so
16:09 RealBadAngel    so at certain angles you can see depth flickering
16:09 RealBadAngel    and its not acceptable
16:10 Taoki[laptop]   RealBadAngel: http://irrlicht.sourceforge.net/forum/viewtopic.php?f=4&t=48992 Might be relevant, not sure if it helps
16:10 Taoki[laptop]   We should ask the Irrlicht people about attributes
16:10 RealBadAngel    ive read it already
16:11 RealBadAngel    its a branch
16:11 RealBadAngel    not official irrlicht
16:11 Taoki[laptop]   ok
16:11 RealBadAngel    we cannot use it at all
16:11 Taoki[laptop]   Hope your latest parallax code is public anyway. I'd like to try it sometime even with the glitching
16:13 Taoki[laptop]   RealBadAngel: Why won't it work with uniform and each instance of a mesh though? Can't that be used to do it per-triangle?
16:13 nore            RealBadAngel, do you know when will that entity duplication bug be fixed?
16:14 * Taoki[laptop] is sorta starting to wish MineTest could switch to raw OpenGL, seeing how problematic Irrlicht always tends to be
16:18 RealBadAngel    tangent vector is calculated for each surface, tile
16:19 RealBadAngel    its not a global variable
16:21 Taoki[laptop]   RealBadAngel: ok. Anyway, if you still want please look at the colored fog too. IIRC you were pending to look at it before including it
16:22 RealBadAngel    Taoki[laptop], once i solve tangent problem we will be able to get any shaders we want to
16:22 RealBadAngel    including rainbow shadows
16:22 Taoki[laptop]   Fog color isn't a shader, all Minetest code. I prefer it that way also
16:23 Taoki[laptop]   Just saying it's unrelatefd
16:23 RealBadAngel    i have to many things started but not finished,  i decided now i wont touch anything till i solve this problem
16:23 RealBadAngel    once and for all
16:23 Taoki[laptop]   RealBadAngel: Must you (as in the MineTest devs) solve the problem, or must Irrlicht do it first?
16:23 Taoki[laptop]   Ok, fair enough
16:24 RealBadAngel    *too
16:25 RealBadAngel    im pretty sure that irrlicht devs should provide attribute usage
16:25 Taoki[laptop]   I shall ask on their forum and IRC about that
16:25 RealBadAngel    but for now im forced to look for another (even hacky) ways to achieve it
16:25 Taoki[laptop]   Least I can help with too
16:25 RealBadAngel    if youre familar with those guys please do so
16:26 RealBadAngel    lack of attribute is pain in the ass
16:26 PilzAdam        Taoki[laptop], if irrlicht fixes it now, then we will probably be able to use it in 5 to 10 years
16:26 RealBadAngel    hehe true
16:26 Taoki[laptop]   PilzAdam: I see what you mean, if you're referring to having to keep compatibility with current Irrlicht versions for that amount
16:27 Taoki[laptop]   ALTHOUGH, we could make features like parallax dependent on Irrlicht version, if they require new fixes there
16:27 Taoki[laptop]   Prolly works with an #ifdef tag
17:22 proller         https://github.com/proller/minetest/compare/weather_fog
19:18 Taoki[laptop]   RealBadAngel: Got confirmation that attributes are not supported in Irrlicht 1.7 and 1.8 at all. So you'll likely have to find another way sadly
19:18 proller         ogre or sdl ? ;)
19:18 Taoki[laptop]   yeah, good question at this point :)
19:19 Taoki[laptop]   Kinda upset at OGRE too because their shaders don't seem to work for ATI + Linux
19:19 Taoki[laptop]   But we ATI Linux user are used to getting some discrimination from the software xD
19:20 sapier          guys switching 3d engine is a very very big task  it's useless to restart the discussion every 4 weeks without anyone who is willing to spend at least 2-4 weeks full time work
19:21 Exio4           *at least*
19:21 Taoki[laptop]   Still, I wonder how hard it would be to switch to raw openGL at this point. Probably pretty hard, but stil
19:21 Taoki[laptop]   Ah, that actually answers that lol
19:21 Exio4           at least you mean for getting a basic minetest that half-works?
19:21 sapier          switching to raw opengl is a half year job
19:21 sapier          full time of course
19:21 Taoki[laptop]   ouch
19:22 Taoki[laptop]   Hmm... I wonder something else however. But I doubt it's stable and possible...
19:22 sapier          about 1-2 months for gui toolkit rewrite and same time for rest of 3d environment (at least)
19:22 Exio4           c55 said something about "the minetest rewrite"
19:22 Exio4           if it is ever done, there is something more to it!
19:23 Taoki[laptop]   Would it be possible to compile MineTest with custom Irrlicht patches? For examples, there is a patch that gets Instancing working under Irrlicht. Perhaps Minetest could include it with directives to compile Irrlicht using that code replacement?
19:23 sapier          yes that'd be not much more yes
19:23 sapier          network code scripting api map code (except formspec) could be preserved
19:24 sapier          so just forget about it for now ;-)
19:24 Taoki[laptop]   What about my idea? Can MineTest include Irrlicht patches and compile with those??
19:24 PilzAdam        why forget it? it would be nice if somone would do it
19:24 sapier          pa i said "for now"
19:25 sapier          if someone is willing to do that of course ;-)
19:25 Taoki[laptop]   PilzAdam: Yeah, but who has the knowledge + time + energy to?
19:25 sapier          I don't see a way to patch irrlicht which is part of users distribution
19:25 sapier          unless we include irrlicht to minetest I don't think we can add own patches
19:25 PilzAdam        Taoki[laptop], I guess none of the core devs
19:26 Taoki[laptop]   Doubtfully, sadlly
19:26 sapier          taoki major problem is time
19:26 Taoki[laptop]   sapier: Some projects choose to come with their own version of a rendering engine, though that's harder to maintain. Grit Engine for instance uses its own version of OGRE
19:27 Taoki[laptop]   sapier: Ironically, I have the time but now the knowledge and energy
19:27 sapier          and it's not a thankfull job at all ... after you switched you'll have to fix incompatibilities ... all that time you hear "but it did work before" ;-)
19:27 Exio4           i guess more than time the main problem is the pain it is
19:29 Taoki[laptop]   Can't say Irrlicht is all bad, but it occasionally has disappointing limitations you wouldn't expect
19:30 sapier          I'm not sure if we didn't just switch irrlicht limitations against xxx limitations
19:31 Taoki[laptop]   Raw OpenGL prolly wouldn't have any limitations except hardware ones. OGRE would have fewer limitations than Irrlicht I believe, but can't say for sure
19:32 sapier          raw opengl has limitation of missing gui as well as missing support tools ...
19:33 sapier          you can do everything in assembler too still for most things c/c++ (or any other high level language) is better ;-)
19:33 Taoki[laptop]   Yeah, those we'd have to make on our own
19:33 Taoki[laptop]   I wonder if it would ever be a choice to have our own version of Irrlicht? But repasinb it on Irrlicht updates would be hard
19:33 Taoki[laptop]   **rebasing
19:33 sapier          I'm not willing to write a own 3d engine ;-)
19:35 Taoki[laptop]   PilzAdam: What do you think about a MineTest specific version of Irrlicht? Would that be a bad idea, and possible to maintain updated with originall Irrlicht?
19:36 Taoki[laptop]   Such could easily work around the current issues... like difficult-to-achieve dynamic lighting, lack of shader attributes, etc
19:36 Taoki[laptop]   Not sure if there are enough reasons for this yet though
19:37 Taoki[laptop]   It could also include any improvements that original Irrlicht doesn't choose to. That goes back to the VBO code issue
19:38 Taoki[laptop]   Aaaahhh... I just remembered the VBO code isn't in yet >:O Ahem... but yeah
19:38 kahrl           the VBO problem could be fixed in minetest, no need to touch irrlicht (... I think)
19:38 Exio4           yeah
19:38 Exio4           there was an idea that hmmmm suggested
19:39 Exio4           copying the "chunk behavior" of minecraft, that is, load a range of X mapblocks around the player and unload the ones that are far away
19:40 sapier          exio as far as I know it's already done that way
19:41 kahrl           no, the mapblocks that are out of the viewing range are still in memory
19:41 kahrl           including their meshes (which is causing the VBO troubles)
19:41 sapier          when are those unloaded?
19:41 Exio4           but generating the meshes is slow right now
19:41 Exio4           sapier: after 10 minutes (default setting)
19:42 kahrl           client_unload_unused_data_timeout
19:42 Taoki[laptop]   Yeah, 10 minutes is too long which is the problem
19:42 sapier          so it's just a big cache ... what about switching from time to memory based exceeding?
19:42 kahrl           but setting the timeout lower causes more network traffic which is undesirable
19:43 kahrl           right now there is some code that relies on the assumption that every loaded mapblock also has a mesh
19:43 sapier          kahrl why not remove time at all and only unload if something new is loaded and some memory limit is exceeded?
19:44 kahrl           guess it could be done that way
19:45 sapier          I don't like timeouts because they tend do be wrong all the time ;-)
19:45 kahrl           you need some way to track the memory used by a mapblock
19:45 sapier          yes do we have a way to do that?
19:45 kahrl           no
19:46 sapier          difficult to add?
19:46 kahrl           which is why when I implemented max_clearobjects_extra_loaded_blocks I defined it as number of blocks and not amount of bytes
19:46 kahrl           not sure how difficult it is
19:46 kahrl           I doubt irrlicht gives you way to get the memory used by a VBO, but it could be estimated
19:47 kahrl           s/you/you a
19:47 sapier          I guess this could fix a lot of issue so it may be worth to investigate that issue
19:51 Exio4           i guess a mapblock limit (if there is some 'aproximated value') would be better anyway
19:52 sapier          yes mapblock limit is still better then time limit
19:53 Exio4           memory usage can be guessed but it can't be accurate
19:54 PilzAdam        hmmmm, I wouldnt mind if you change the heat and humidty used for biome selection
19:54 sapier          if we don't have a exact number I'd prefere the mapblock way
19:54 PilzAdam        I can easily adjust the values used in Pilztest