Time  Nick         Message
02:26 RealBadAngel any more comments on #3612?
02:26 ShadowBot    https://github.com/minetest/minetest/issues/3612 -- Filmic HDR tone mapping by RealBadAngel
02:49 RealBadAngel !tell est31 i told ya that you cant make enable_animations static. now you have to shutdown the clienat and restart for setting change to work
02:49 ShadowBot    RealBadAngel: O.K.
04:13 RealBadAngel everybody sleeping or what? :)
04:15 RealBadAngel side effect of inventory meshes instead of textures patch, item_image_buttons cant show labels (yet) but easily can show stack count (which was previously impossible)
04:16 RealBadAngel i thought i will have to code that feature but it seems it coded itself lol
04:22 * sofar      back from .nl
04:23 paramat      ok hmmmm #3649 is updated on your good advice and tested
04:23 ShadowBot    https://github.com/minetest/minetest/issues/3649 -- FindSpawnPos: Let mapgens decide what spawn altitude is suitable by paramat
04:23 paramat      welcome back
04:24 hmmmm        maybe I'm misunderstanding the point of this function
04:24 hmmmm        getSpawnLevelAtPoint?
04:25 sofar        paramat: ty. still 10hrs of jet lag to get rid of :)
04:25 hmmmm        you want a function per mapgen to tell you the y level of a point if it is a good place to spawn, otherwise none
04:25 hmmmm        bool Mapgen::isSuitableForSpawn(v2s16 query_pos, int *ground_level)
04:25 hmmmm        int y;
04:26 hmmmm        if (mg->isSuitableForSpawn(v2s16(guess_x, guess_z), &y)) break;
04:26 hmmmm        ?
04:27 hmmmm        "getSpawnLevelAtPoint" doesn't really describe what it is you're after
04:27 hmmmm        wtf is the "spawn level"
04:28 paramat      well, the function returns the spawn y in the column given by findSpawnPos, but if the column is not suitable 'map gen limit' is returned to signify that
04:28 paramat      y of player spawn
04:28 hmmmm        I get that, but I'm saying that it might not be the best choice from an interface standpoint
04:29 paramat      here 'at point' means 'in this node column' just as in 'getGroundLevelAtPoint'
04:29 hmmmm        you're getting two pieces of information here:  is the (x,z) suitable for spawning?  if it is, what is the ground level at this point?
04:29 paramat      correct
04:29 hmmmm        well it's up to you really
04:30 hmmmm        I'm just saying
04:30 paramat      yes i'm using 'map gen limit' as a return value to signify unsuitable spawn, it's nice and simple
04:31 hmmmm        getSpawnLevelAtPoint probably isn't the best name for this
04:31 hmmmm        getSuitableSpawnHeight?
04:31 hmmmm        idk
04:31 paramat      needs to describe the possibility of unsuitable column
04:31 paramat      ok
04:31 hmmmm        no, I don't think my name is any better
04:32 hmmmm        if you can come up with something that describes it better, change it
04:32 paramat      ok
04:32 hmmmm        from a code standpoint it looks fine to me
04:32 hmmmm        you basically copied getGroundLevelAtPoint
04:32 hmmmm        I would add a comment in mapgen.h though
04:33 hmmmm        just mention what this function does, and what its possible return values are
04:33 paramat      ok
04:34 hmmmm        ehh
04:34 hmmmm        was it your intention to remove getGroundLevelAtPoint completly?
04:34 hmmmm        completely*
04:35 paramat      see commit message. we can re-add it in correct form soon
04:35 paramat      it's never used, it was only used to get spawn y
04:35 hmmmm        what about v7?
04:35 hmmmm        that was relatively accurate
04:36 paramat      it was, if you want lowest ground level
04:36 paramat      i guess lowest is what people would want
04:36 hmmmm        getGroundLevelAtPoint was supposed to be exposed to the lua api for sapier's thing but i'm not sure why he never ended up using it
04:37 paramat      how about i make sure to re-add these functions soon?
04:37 paramat      in another commit
04:37 hmmmm        well like i said it's up to you
04:37 paramat      then it can be accessed from a new API
04:37 hmmmm        if they're not being used then nobody will miss them
04:38 paramat      there's also 'get biome at point' never used but modders will love that
04:38 paramat      ok
04:38 hmmmm        yea
04:38 hmmmm        so however you decide to proceed is fine
04:38 hmmmm        just make sure modders and users are happy :)
04:38 hmmmm        +1
04:39 paramat      thanks will update this
04:42 paramat      'getSpawnYAndSuitabilityAtPoint()' ..
04:42 hmmmm        ehh
04:42 hmmmm        just stick with what you have right now and explain it in a comment
04:46 paramat      ok
04:49 RealBadAngel hi guys
04:50 RealBadAngel so, can we have 3612 merged?
04:51 hmmmm        #3612
04:51 ShadowBot    https://github.com/minetest/minetest/issues/3612 -- Filmic HDR tone mapping by RealBadAngel
04:51 hmmmm        doesn't this require IrrPP
04:51 RealBadAngel no
04:51 RealBadAngel thats the only pp effect that can go in with current shaders
04:53 RealBadAngel it just modifies final color
04:55 hmmmm        is there any way to cut down on duplicated shader code?
04:55 RealBadAngel which part do you mean?
04:55 RealBadAngel nodes and water ones?
04:55 hmmmm        the same tonemapping code is repeated in both the node and water shaders
04:56 hmmmm        aren't you able to make a 'common' shader file?
04:56 RealBadAngel water shader in its current form will be removed
04:56 hmmmm        well, whatever
04:56 RealBadAngel i need another commit for it
04:56 RealBadAngel but later on
04:57 RealBadAngel this will eliminate those duplicated code
04:57 hmmmm        does this code break the alpha channel?
04:57 RealBadAngel why do you think so?
04:58 RealBadAngel alpha is preserved, only rgb is changed
04:58 RealBadAngel or at least should be, lemme check
04:58 hmmmm        vec4 col = vec4(color.rgb, gl_Color.a); col *= gl_Color;  --->  vec4 col = vec4(color.rgb * gl_Color.rgb, 1.0);
04:58 hmmmm        this is not the same at all
04:59 RealBadAngel damn, it was before but nerzhul wanted to simplify the code
04:59 RealBadAngel i will fix that right now
05:08 RealBadAngel meanwhile: #3676
05:08 ShadowBot    https://github.com/minetest/minetest/issues/3676 -- Fix displaying item_image_button label by RealBadAngel
05:09 hmmmm        dammit RBA :)
05:09 hmmmm        your 6cd2b3b seems to be causing a lot of regressions
05:09 hmmmm        who signed off on this!?!?
05:11 RealBadAngel this was complex change and lotsa folks were testing it. it happens that sometimes not everything is caught
05:12 RealBadAngel if i got report about it earlier i would code it right away
05:12 RealBadAngel it was 30min work
05:12 hmmmm        mm
05:12 hmmmm        so let me understand this
05:12 RealBadAngel about 3612, youre wrong, everything is fine, alpha is preserved as it should
05:13 hmmmm        what was it about 6cd2b3b that made it regress?
05:13 hmmmm        in 3676 you remove the text from the button and instead superimpose a static text control over it
05:13 RealBadAngel completely new way of displaying items, meshes instead of textures
05:13 sofar        paramat: I'm not worried about the gravel texture needing revisions - I'd rather redo it 5+ more times and get it "right"
05:13 hmmmm        oh I see
05:13 hmmmm        right
05:13 RealBadAngel so buttons had to be emulated
05:14 hmmmm        so now those buttons have no text or graphics on them
05:14 hmmmm        they have 3d meshes drawn in that area
05:14 paramat      ok
05:14 RealBadAngel yes
05:14 hmmmm        and now with this they have text superimposed on top of all that
05:14 hmmmm        that's sensible I'd say
05:14 RealBadAngel same as itemstacks
05:14 RealBadAngel which gained ability to display stack count btw
05:15 RealBadAngel so mods can become easier
05:15 hmmmm        formspecs didn't previously have the ability to display static text controls?
05:15 RealBadAngel https://github.com/minetest-technic/unified_inventory/commit/b3c6c922861f9936cc03fae5cd6cabb9cacd9e27
05:15 RealBadAngel only labels
05:15 RealBadAngel but those have focus
05:15 hmmmm        okay
05:16 RealBadAngel so if you put there a label, you wont be able to push that button
05:16 hmmmm        RBA, I have a complaint about the commit message
05:16 hmmmm        it's pretty vague
05:16 hmmmm        why not make it something like:
05:16 RealBadAngel static text is well, just static ;)
05:16 RealBadAngel that even could be exposed for mods to use
05:17 hmmmm        FormSpec: Add StaticTextSpec and superimpose over item image buttons
05:17 hmmmm        This fixes Issue #3672, "Inventory items are covering item_image_button labels"
05:17 ShadowBot    https://github.com/minetest/minetest/issues/3672 -- Inventory items are covering item_image_button labels since 6cd2b3b
05:17 hmmmm        ?
05:17 hmmmm        how about that
05:17 RealBadAngel i can add new formspec element, thats ok
05:18 hmmmm        it explains what the fix is and how it solves the problem
05:18 RealBadAngel but buttons have to be emulated anywyay
05:18 hmmmm        whereas the current commit message says, "Fix thing"
05:18 RealBadAngel ok, i will explain it more
05:18 hmmmm        it's a step up from proller's "fix" commit messages
05:18 hmmmm        lots of room for improvement though
05:19 RealBadAngel hehe i got that feeling when typing that commit message lmao
05:22 RealBadAngel ok, ive changed it as you suggested
05:23 hmmmm        anyway back to 3612
05:24 hmmmm        so the way it is right now, without your commit, it seems like   vec4 col = vec4(color.rgb, alpha);  col *= gl_Color;   is not functionally the same as the code you replaced it with
05:25 hmmmm        that being said, the current code is the thing that seems faulty, and your new code is correct
05:25 hmmmm        because look
05:25 hmmmm        with the old one you set col to color.rgb, gl_Color.a
05:25 hmmmm        then you multiply col by the entire gl_Color, not gl_Color.rgb
05:26 hmmmm        so that means col.a = gl_Color.a^2
05:27 hmmmm        but then at the end after all the calculations, the alpha is overridden again
05:27 hmmmm        this only matters if the alpha channel plays a role in the intermediate computation of any of the RGB values
05:27 RealBadAngel hold on
05:27 RealBadAngel http://pastebin.com/aEkVsYiM
05:27 RealBadAngel please read that piece
05:28 hmmmm        ?  yes I know that already
05:28 RealBadAngel ive put there explanations whats going on here
05:28 hmmmm        I am just saying that the new code and old code might not create the same exact result, if the alpha channel is used in calculating rgb anywhere else
05:29 hmmmm        which it doesn't seem so
05:29 RealBadAngel its not, cant be
05:29 hmmmm        I'm not saying your new code is wrong
05:30 RealBadAngel maybe before it wasnt made clear that alpha doesnt play any role and is restored at the end
05:30 hmmmm        okay whatever this is settled
05:30 hmmmm        no changes need to be made
05:30 hmmmm        what about the #define ENABLE_TONE_MAPPING?
05:31 RealBadAngel what about it?
05:31 hmmmm        I added notes to the PR
05:31 hmmmm        but I asked if you need to #define it to a value of 1 in order to activate
05:33 RealBadAngel checkin it now
05:34 hmmmm        you might be able to optimize tonemapping by precalculating whiteScale
05:34 RealBadAngel well, ENABLE_BUMPMAPPING also is not using 0/1 and i dont recall any reports that bumpmapping is not working on ATI
05:34 hmmmm        what is the reason for some of the defines being 0/1 and some others not, then?
05:35 RealBadAngel frankly i cant remember right now, its old code
05:35 hmmmm        mmmmmmmmmmmmmmmm
05:35 hmmmm        maybe you should've added COMMENTS
05:35 RealBadAngel it was something about complex IFs on compilation time
05:36 hmmmm        if you do remember, please make a PR that adds a comment explaining why
05:36 RealBadAngel if you check only if variable exists, you dont need to check for its value
05:36 RealBadAngel but if you want to evaluate its value using and, or you need its value
05:37 hmmmm        well, duh
05:37 RealBadAngel so, its enough when you check: #ifdef ENABLE_PARALLAX_OCCLUSION
05:38 RealBadAngel you dont need the value, but fact its here
05:38 RealBadAngel if (GENERATE_NORMALMAPS == 1 && normalTexturePresent == false)
05:38 RealBadAngel here you need its value
05:38 hmmmm        yes but i thought you said that certain shader compilers interpret it in different ways
05:38 RealBadAngel no
05:38 RealBadAngel look, theres #ifdef
05:38 RealBadAngel if defined
05:38 hmmmm        okay
05:39 RealBadAngel so #ifdef ENABLE_TONE_MAPPING is absolutely correct
05:39 hmmmm        great
05:41 RealBadAngel about optimization, i could precompute some things to make it faster
05:41 RealBadAngel and leave original formulas as comments
05:42 hmmmm        yeah, that's what I would do for whiteScale
05:43 RealBadAngel gain wont be huge, but here all little things do accumulate
05:43 RealBadAngel its calculated per each pixel
05:49 RealBadAngel in fact i can precalculate more things in main equation
05:53 RealBadAngel 3 muls, 1 div less per one execution
05:55 RealBadAngel well, div is 0.01 / 0.3 im not sure if i can put there precalculated one
05:55 RealBadAngel hmmmm, what do you think?
05:56 RealBadAngel its 0,03(3)
05:56 hmmmm        E / F?
05:56 RealBadAngel yeah
05:56 hmmmm        I don't think combining those constants will help
05:56 hmmmm        they look like easy pickins for the shader optimizer
05:57 RealBadAngel they do modify the colors curve
05:57 RealBadAngel im not sure we do want to make them settable by user
05:58 RealBadAngel you mean optimizing by compiler?
05:58 hmmmm        yes
05:59 RealBadAngel i have no idea if those get optimized
05:59 RealBadAngel it may depend on drivers
06:00 hmmmm        to represent 0.03(3) what you should do is round it to 0.03334
06:00 hmmmm        floats typically have 5 decimal places of accuracy
06:00 RealBadAngel ok, will try
06:00 RealBadAngel also im replacing C*B, D*E and D*F
06:01 RealBadAngel surely it wont harm if we do optimize it on our own
06:01 hmmmm        sure
06:16 RealBadAngel hmmmm, something like that: http://pastie.org/10714316
06:25 hmmmm        yes
06:25 hmmmm        looks good
07:07 paramat      ok kahrl hmmmm i'll merge #3661 in a moment along with #3649
07:07 ShadowBot    https://github.com/minetest/minetest/issues/3661 -- Remove ClientMap::m_camera_mutex by kahrl
07:07 ShadowBot    https://github.com/minetest/minetest/issues/3649 -- FindSpawnPos: Let mapgens decide what spawn altitude is suitable by paramat
07:10 paramat      now merging
07:24 paramat      merged
07:48 RealBadAngel hmmmm, updated 3612 with optimizations
07:50 hmmmm        looks good, but you never answered that one note
07:50 hmmmm        shouldn't exposureBias be const float?
07:51 RealBadAngel ah, indeed
07:53 RealBadAngel done
08:01 RealBadAngel hmmmm, so any ohter comments on 3612 or 3676?
08:02 hmmmm        3612 is merged
08:03 hmmmm        what's 3676 again??
08:03 hmmmm        oh that
08:04 RealBadAngel whoever can confirm its working its me, it hit UnifiedInventory
08:04 RealBadAngel ive already updated UI to reflect latest changes
08:05 hmmmm        i can't test it
08:05 RealBadAngel afaik UI is the only mod using that feature so its propably only one affected
08:06 hmmmm        vanessae's server uses UI right?
08:06 RealBadAngel yeah
08:06 hmmmm        i'd try that out but it'd take longer to connect to it than it would to go through the tedious process of going to the forum, searching for the mod, downloading it, extracting it, making a new world with it enabled, etc.
08:06 hmmmm        ugh
08:07 RealBadAngel just grab UI
08:07 hmmmm        well it doesn't matter
08:07 RealBadAngel https://github.com/minetest-technic/unified_inventory
08:07 hmmmm        i'm sure if it doens't work somebody else will complain
08:09 RealBadAngel ok, so now im gonna finish #3166
08:09 ShadowBot    https://github.com/minetest/minetest/issues/3166 -- Send to clients only changed node metadata instad of whole mapblock by RealBadAngel
08:09 RealBadAngel it waited for too long already
08:09 hmmmm        what does it still need?
08:10 RealBadAngel a) elimination od double meta updates per tick
08:10 RealBadAngel idk yet a reason for those, need to find out
08:10 RealBadAngel b) gather all the metas and send them as single package
08:11 RealBadAngel before b) was kinda done by sending whole mapblock, but thats unacceptable anymore
08:12 RealBadAngel sending many small packages is better than whole block already, but better would be a single one, dont you agree?
08:17 kahrl        RealBadAngel, slight regression with item image buttons: previously the item and label moved a few pixels when pressing the button (giving a 3d appearance), now neither moves
08:18 RealBadAngel kahrl, that i cant emulate
08:18 RealBadAngel buttons are drawn by irrlicht, images and text by formspec code
08:18 kahrl        yeah I'm not sure if it is actually needed
08:19 kahrl        though it would be nice to have, if only for consistency with normal buttons
08:19 RealBadAngel item image buttons are already very specific and i dont think they should act the same
08:20 RealBadAngel at least thats not needed
08:20 RealBadAngel pressing the button effect is still here
08:21 RealBadAngel but to keep that consistency patch for irrlicht would be needed not for formspec
08:24 kahrl        I suppose you could check button->isPressed() when rendering the item and label
08:26 RealBadAngel shitload of things should be rewritten then
08:26 RealBadAngel we dont render formspecs each frame
08:27 RealBadAngel im not sure thats doable even
08:27 RealBadAngel not "render" but "regenerate"
08:29 kahrl        oh, I thought I saw a loop over the StaticTextSpecs that did a m_font->draw
08:32 RealBadAngel checkin now it could be possible to offset it when drawing text
08:47 RealBadAngel fuck
08:47 RealBadAngel i can code it, hmmmm is gonna kill me ;)
08:48 RealBadAngel kahrl, ^^
08:51 kahrl        RealBadAngel: heh
08:52 RealBadAngel but i can do that easily for labels on the buttons
08:52 RealBadAngel in fact i coded that already
08:52 RealBadAngel another thing are images
08:52 kahrl        images should be similar, no?
08:53 RealBadAngel but those are another story, theyre in HUD
08:53 RealBadAngel nah, that will be easy too
08:53 kahrl        aren't they ImageDrawSpecs, in GUIFormSpecMenu?
08:53 RealBadAngel exactly
08:54 RealBadAngel ok i know what to do
08:54 RealBadAngel talking really helps
08:55 RealBadAngel while im still back there, maybe i will add static_text to formspecs?
08:55 RealBadAngel that is really useable
08:55 RealBadAngel formspecs doesnt have that feature exposed
08:55 kahrl        hmm, what would be the difference to label?
08:55 RealBadAngel not being able to hover
08:56 RealBadAngel labels are stealing focus
08:56 kahrl        huh
08:56 kahrl        why would a label be able to get focus
08:56 RealBadAngel put a label that will cover a button
08:57 RealBadAngel you wont be able to press that button
08:57 kahrl        oh
08:58 RealBadAngel so in some cases static text would be really better than labels
08:59 kahrl        I think from an interface point of view it shouldn't be a totally new kind of formspec element, but an option to label
08:59 RealBadAngel or that
08:59 RealBadAngel even easier to code
08:59 Hijiri       clearly formspecs need to be an FRP framework
09:01 RealBadAngel oh we had many talks what formspecs should be, easier to talk about than code it ;)
09:01 RealBadAngel at some point SN was coding them as lua tables i think
09:07 RealBadAngel kahrl, ok, made those labels on button clickin, now images
09:08 kahrl        nice
09:08 nrzkt        RealBadAngel, isn't this better to add a label attribute to make them static text ?
09:09 RealBadAngel nrzkt, thats exactly what we have agreed on
09:09 RealBadAngel but i will do that in separate commit
09:40 everamzah    names are behind the water.  (water is obstructed view of names)
09:43 everamzah    http://imgur.com/a/TYSun
09:45 RealBadAngel kahrl, #3678
09:46 ShadowBot    https://github.com/minetest/minetest/issues/3678 -- Formspec: recreate item_image_button pressed state for its image and … by RealBadAngel
09:46 RealBadAngel can you test it?
09:57 kahrl        RealBadAngel: instead of the hardcoded values I'd suggest using skin->getSize(gui::EGDS_BUTTON_PRESSED_IMAGE_OFFSET_X) and _Y
09:57 kahrl        for the image
09:57 kahrl        and for the text, skin->getSize(gui::EGDS_BUTTON_PRESSED_TEXT_OFFSET_X) and _Y
09:59 kahrl        maybe call the field "parent_button" instead of "button, to make the intent clearer
09:59 RealBadAngel ok will try them
09:59 kahrl        "button"*
09:59 RealBadAngel sure
10:01 kahrl        also it seems like line 1506 through 1509 are no longer needed at all
10:06 RealBadAngel indeed
10:07 kahrl        hooray for less code :)
10:09 RealBadAngel at least at some spots ;)
10:10 kahrl        touche :P
10:15 everamzah    sometimes this halo(?) thing doesn't update http://imgur.com/a/2NcBW
10:18 RealBadAngel doesnt update? what do you mean?
10:18 everamzah    sometimes it's not see-through.  but opaque, rather, and i can't see at all what i'm pointing at
10:19 everamzah    normal: http://imgur.com/jjLLTWJ
10:19 everamzah    but maybe it's normal, because sometimes that's how it is for see-thru nodes and water
10:21 RealBadAngel are you using noclip when it becomes opaque?
10:21 everamzah    no
10:21 RealBadAngel please can you determine when exactly it happens?
10:22 everamzah    no.  kind of a rare thing.  _felt_ like it wasn't updating, but i'm sorry i can't be clearer
10:22 everamzah    it's probably nothin :O
10:22 RealBadAngel shouldnt happen at all
10:22 RealBadAngel i will try to look for a possible reason
10:22 everamzah    diggin the new gfx changes, tryin some em out now
10:22 everamzah    maybe some1 else will notice it, too
10:22 RealBadAngel for sure
10:23 * everamzah  wonders if it's a graphics driver/memory thingy on his debian laptop
10:23 RealBadAngel its a simple mesh
10:23 RealBadAngel but a change to clean it up was rather huge
10:23 RealBadAngel over 20 files touched
10:23 everamzah    \o/
10:24 RealBadAngel so its possible that somwhere sits a nasty bug and just pops up from time to time ;)
10:25 everamzah    i don't notice it now.  but before there was another player nearby and it seemed to happen more (?)
10:26 everamzah    i wonder why these ones are offset: http://imgur.com/a/4ysFc
10:26 everamzah    maybe it's the node selection size def
10:27 kahrl        everamzah: that's because it scales the mesh outwards from the center of the node
10:27 everamzah    ah
10:28 RealBadAngel everamzah, can you make a screenshot of that chair with boxes selection mode?
10:28 everamzah    sure
10:29 * everamzah  scrolls furiously
10:30 everamzah    sry, offhand do u know the setting name? (haha)
10:30 RealBadAngel use menu
10:30 everamzah    ya i am i'm scrolling.  i forgot what i turned on, even
10:30 everamzah    oh
10:31 everamzah    node_highlighting i found it
10:31 * everamzah  crosses fingers for a search box
10:31 RealBadAngel its in client/graphics/ingame/basic
10:32 everamzah    http://imgur.com/a/rrEa0
10:32 everamzah    looks tight, on the box
10:32 everamzah    afk
10:34 RealBadAngel ok, i will try to correct the halo then
10:34 RealBadAngel kahrl, ok, ive updated the PR
10:37 kahrl        RealBadAngel: code looks good to me now, will test
10:38 kahrl        oh, git says 1 line adds whitespace errors
10:39 kahrl        I've marked it on github
10:40 kahrl        and s/button/parent_button in guiFormSpecMenu.h:251
10:42 RealBadAngel ok
10:51 everamzah    seems to be opaque pretty often when there's two players together
10:52 * everamzah  wonders whether you saw the imgur album first linked with the names behind water
10:54 RealBadAngel thats pretty normal
10:54 everamzah    oic
10:54 RealBadAngel player name tags are rendered wrong way
10:54 RealBadAngel that will be changed when i move tags to HUD where they should be
10:55 RealBadAngel those tags cannot be rendered as 3d obects in the world, i have to move them away from here
10:56 RealBadAngel otherwise tags will be shaded with all the postprocessing effects
10:56 RealBadAngel so its on my TODO list already
10:59 RealBadAngel kahrl, ok. fixed. does the change work for you?
11:03 kahrl        RealBadAngel: still a compile error :(
11:03 kahrl        I've marked the two bad lines
11:03 RealBadAngel wut?
11:03 kahrl        did it compile for you?
11:03 RealBadAngel will check clean sources
11:03 RealBadAngel maybe ive messed something
11:04 kahrl        error: expected ‘,’ or ‘;’ before ‘rect’
11:04 kahrl        in line 2440 and 2479
11:04 RealBadAngel lol yeah
11:05 RealBadAngel looks like i was coding on other folder
11:05 RealBadAngel and compiling another one
11:08 RealBadAngel gui::EGDS_BUTTON_PRESSED_TEXT_OFFSET_X seems to be always 0
11:08 RealBadAngel only Y changes
11:09 RealBadAngel for it to be nice lookin i think i shall use (Y, Y)
11:10 kahrl        or just use image offset for both
11:10 kahrl        that's more correct imho
11:10 RealBadAngel what about different fonts?
11:10 RealBadAngel some big ones? offset propably will be then smaller for text
11:11 RealBadAngel Y, Y works ok
11:11 kahrl        I don't think the offset should depend on font size
11:11 RealBadAngel we dont know that
11:11 kahrl        yeah but using Y for X seems very wrong
11:12 RealBadAngel can you try image one?
11:13 kahrl        works well
11:13 RealBadAngel image one offset seems to be smaller indeed
11:13 RealBadAngel so leave there IMAGE one?
11:13 kahrl        yeah
11:13 RealBadAngel ok
11:13 kahrl        but write a small comment why it's IMAGE and not TEXT
11:19 RealBadAngel done
11:19 RealBadAngel seems to be polished now
11:22 kahrl        lgtm now
11:51 Megaf        Hi all, can you test this for me? Start a server (with pre generated map), wait 10 minutes, observe memory use on the server. Join the server, walk around in your already generated map, for about 10 minutes, observe the memory use climb. Leave the server, wait 10 minutes, memory use will not return to when you started the server. Memory use only climbs.
11:51 Megaf        no matter the settings in minetest.conf
11:52 Megaf        in my case it climbed from around 112 MB to 192 MB
11:52 Megaf        193*
11:53 nrzkt        Megaf, Player* objects are not free, only session & playerSAO
11:56 Megaf        I don't understand that
13:50 xeranas      Hello, minetest.add_particlespawner have parameter colissiondetection and I wonder if I can destroy particle on its colission?
15:09 sofar        xeranas: you can't right now, no
16:25 RealBadAngel wb hmmmm
17:04 RealBadAngel kahrl, hmmm, here?
17:39 Fixer        hmhmhm, got craaash on latest git playing on 0.4.12 just test server, logs are clear
17:43 Fixer        nt.dll error
17:59 red-001      that happens to me sometimes
17:59 red-001      usually when quitting the game after it was running for a long time
18:00 Obani        RealBadAngel, does your master branch on Minetest-HD contain the dof ?
18:03 RealBadAngel no
18:03 RealBadAngel postprocessing is far from being complete
18:04 Obani        RealBadAngel, do you have screenshots of dof in action, I can't recover them in my history
18:04 Obani        ?
18:07 red-001      https://www.google.ie/search?q=site%3Airc.minetest.ru+inurl%3Aminetest-dev+dof
18:07 RealBadAngel http://i.imgur.com/dCfPFYq.png
18:08 Obani        Thanks RealBadAngel
18:09 sofar        RealBadAngel: I must say, DoF+tonemapping is pretty
18:09 sofar        RealBadAngel: did you ever look at those shader compile errors ?
18:09 RealBadAngel what errors?
18:11 sofar        I posted those ... 5/6 days ago?
18:11 RealBadAngel i was fixing some for sure
18:12 sofar        it was on my Intel IVB gfx
18:12 RealBadAngel but im sure if the ones you mean
18:13 RealBadAngel anyway i had to step back to root and eliminate obstacles first
18:13 sofar        RealBadAngel: http://irc.minetest.ru/minetest-dev/2016-02-03#i_4523129
18:13 RealBadAngel i will go back to pp soon
18:13 red-001      did you add the fix for building in VS2013?
18:13 sofar        yeah no worries, I was travelling and in CET and didn't see you online at the time
18:14 RealBadAngel red-001, as i said, i will be back to coding pp soon
18:14 RealBadAngel just a few more things to fix first
18:15 RealBadAngel i need clean depth buffer, without any garbage here first
18:15 sofar        RealBadAngel: also DoF gave me some weird black sufaces in some cases, I wonder if that was parralax occlusion
18:19 RealBadAngel sofar, that was bumpmapping given wrong textures
18:19 RealBadAngel theres shitload of things to fix
18:19 RealBadAngel but one by one
18:19 sofar        cool
18:21 RealBadAngel if i will start writing now whats need to be fixed you would sit there and read for half an hour
18:21 RealBadAngel so just be patient ;)
18:22 RealBadAngel also, from fixing old thing quite nice new features pops in
18:22 RealBadAngel see latest commits
18:22 sofar        I'll try and test on Intel later this week again
18:23 RealBadAngel sofar, btw, you should press your commits waiting
18:23 sofar        I've been out for a week, and I'm patient
18:23 sofar        but yes, there's a few that really are getting stale at this point
18:24 RealBadAngel why 3503 is idle?
18:25 RealBadAngel move your ass and finish it
18:26 sofar        honestly I don't understand why it wasn't merged ~ 16 days ago
18:26 hmmmm        #3503
18:26 ShadowBot    https://github.com/minetest/minetest/issues/3503 -- Nodebox: Allow nodeboxes to "connect" by sofar
18:27 sofar        the cleanups requested don't necessarily make the code better
18:27 hmmmm        it has a "WIP" tag
18:27 sofar        I can't remove tags
18:27 hmmmm        i see.
18:27 sofar        est31 already +1'd and said it could get merged as is
18:28 sofar        I agree it's not ideal code, but due to the fact that there's no common class inheritance this is what it is
18:28 sofar        also, it's fairly simple code, really
18:28 hmmmm        well things usually require at least two code review +1s
18:28 sofar        hmmmm: a review from you would be very helpful ;)
18:29 hmmmm        (note that there are some non-developers who take it upon themselves to +1 a PR that they didn't actually review, they just like the concept of)
18:29 hmmmm        if the code can be made cleaner and it's pretty easy to do so, then please...
18:29 sofar        nrzkt hasn't +1'd it. His reviews have been thorough and both him and est31 can't yet agree on a proper resolution of the non-common-class problem they'd additionally like to see solved
18:30 hmmmm        minetest upstream is meant for mostly production-ready code
18:30 hmmmm        ehh
18:30 hmmmm        no offense to nrkzt, but his reviews tend to be:  "make this variable const"
18:30 sofar        hmmmm: that's the problem, ideally VManip and map are made into related classes so inheritance works for us
18:30 hmmmm        i wouldn't give that as much weight personally
18:30 hmmmm        i'll check it out
18:31 sofar        I'll take this PR to the end, as far as it needs to go, obviously
18:34 RealBadAngel hmmmm, can you also check #3678 ? you shouldnt have much work with it this time
18:34 ShadowBot    https://github.com/minetest/minetest/issues/3678 -- Formspec: recreate item_image_button pressed state for its image and … by RealBadAngel
18:34 hmmmm        one thing at a time
18:34 hmmmm        this sort of pileup is exactly what discourages me from wanting to be active in #minetest-dev
18:34 RealBadAngel sure, just a reminder ;)
18:34 hmmmm        i haven't even been able to code anything myself
18:35 RealBadAngel maybe seeing us on code streak will pump you too?
18:35 hmmmm        I have a full time job too
18:36 RealBadAngel like everybody
18:37 RealBadAngel but you have to admit that mt needs some activity, lately it was rather stagnating
18:37 RealBadAngel over half a year since last release...
18:37 hmmmm        this is what always happens
18:37 hmmmm        if i take a break, minetest stagnates
18:38 RealBadAngel hehe
18:38 nrzkt        hmmmm, i have a reduced activity these days and i mainly look at code robustness where i can't look at code general design
18:38 hmmmm        nothing ever gets done, PRs pile up, and there's an unexplainable performance regression
18:38 RealBadAngel aw cmon, ive solved a few old issues lately
18:38 red-001      nrzkt are you nerzhul?
18:38 nrzkt        RealBadAngel, 0AD also don't have a great cycle, STK too, and that's not a problem :)
18:38 nrzkt        red-001, yes
18:38 hmmmm        i am talking in general that's how the pattern seems to be
18:39 hmmmm        whenever I take a break from minetest and come back, there's a new major performance regression
18:39 sofar        it's unavoidable. Question is how to solve it.
18:39 nrzkt        hmmmm, we didn't see you since some time, and we didn't see a performance regression, ? :)
18:39 hmmmm        maybe that's because i'm not officially "back" yet
18:40 RealBadAngel maybe it could be a good time to get VBO in for a change?
18:40 hmmmm        why is vbo not in
18:40 RealBadAngel because nobody dared?
18:40 hmmmm        ridiculous
18:40 RealBadAngel i was lookin at memory usage today
18:41 RealBadAngel the eating memory is not within VBO itself
18:41 hmmmm        I'll review it after these two pending reviews and then we'll commit
18:41 hmmmm        i know, it's irrlicht
18:41 RealBadAngel no its mt
18:41 hmmmm        what's going on with mt?
18:41 RealBadAngel some blocks get never unloaded
18:41 RealBadAngel i can tell you step by step how to notice it
18:42 RealBadAngel i started to read code around but im not familiar with it
18:43 RealBadAngel but i put there some dumps runtime and checked that certain areas are NEVER unloaded
18:43 * red-001    never sees areas unload
18:44 RealBadAngel so, hardware buffers cant be freed neither
18:45 RealBadAngel check this: spawn at ground level, make sure that you have some are loaded around you, best look at minimap and make sure all around you gets generated
18:45 RealBadAngel now fly up, like y=1000
18:46 RealBadAngel stay at this level for a while
18:46 RealBadAngel look down and press R, you will see the area down there
18:47 RealBadAngel still here
18:47 RealBadAngel teleport at 0,5000,0
18:47 RealBadAngel wait again
18:47 RealBadAngel teleport back to 0,1000,0
18:47 RealBadAngel look down with R
18:47 RealBadAngel area is STILL here
18:48 RealBadAngel no matter what you will do, it will never get unloaded
18:50 red-001      can #3656 be reopened ?
18:50 ShadowBot    https://github.com/minetest/minetest/issues/3656 -- Stop water from spreading over water. by red-001
18:50 red-001      I added a fix for the problem with rivers
18:51 red-001      or should I make a new PR?
18:52 RealBadAngel so, either those blocks are not checked for last usage time once they felt out of range, or somehow their timers are not updated anymore
18:53 RealBadAngel i got a feeling that this can be related to code that allows to keep blocks loaded without active player around
18:54 RealBadAngel because it keeps a single area loaded all the time
18:54 RealBadAngel hope this observations will help some1 finding the reason
18:56 RealBadAngel (note that when you fly up new blocks keep loading and unloading all the time, so it works in general)
19:01 RealBadAngel ok, enough on that topic. i do have a question: i need to resize mesh, kinda scale but not by multiplying vertices by a factor (that works well only for cuboid having (0,0,) as center)
19:02 RealBadAngel so i need to sub  factor from box.min and add to box.max
19:03 RealBadAngel question is, how to call that method? its not scaling. resizing? expanding?
19:03 sofar        is it a scaling with offset?
19:03 RealBadAngel offset is for translation, when you move the box
19:04 RealBadAngel i want to make box bigger by given factor
19:04 sofar        yes, but a normal scale implies that the center remains identical. don't you just scale with a different center?
19:04 RealBadAngel i want to keep the center
19:05 red-001      !seen paramat
19:05 ShadowBot    red-001: I saw paramat in #minetest-dev 11 hours, 40 minutes, and 37 seconds ago saying "merged"
19:05 RealBadAngel and normal scale works only with center at 0,0,0
19:06 RealBadAngel sofar, http://i.imgur.com/YbNphtr.png
19:06 RealBadAngel this is done with scale mesh
19:06 sofar        that helps, a lot :)
19:06 sofar        grow?
19:07 sofar        seems much more sensible as a name
19:07 RealBadAngel factor can be negative too
19:07 sofar        yes, growing can mean downwards too
19:07 sofar        trees grow roots downwards
19:07 sofar        branches get thicker downwards as well as upwards
19:08 RealBadAngel wouldnt expand be better?
19:08 celeron55    you obviously first translate it so that origin is at center, then scale it, and then revert the translation
19:09 celeron55    same thing as if you wanted to rotate it around a point that is not the origin
19:09 sofar        RealBadAngel: expand, sure, that could work as well
19:10 celeron55    so, it's just scaling around a different origin
19:11 RealBadAngel celeron55, that will work fine i think
19:12 RealBadAngel but sub/add (i need small resize just) could be less costly
19:13 RealBadAngel celeron55, http://pastebin.com/wfwuPSyb something like that
19:13 RealBadAngel less code than 2 translations + scale
19:18 red-001      !tell paramat could you reopen https://github.com/minetest/minetest/pull/3656 ?
19:18 ShadowBot    red-001: O.K.
19:20 nrzkt        RealBadAngel, there is a timeout on loaded areas if i remember
19:20 RealBadAngel yes there is
19:20 RealBadAngel problem is that those block get never unloaded
20:59 nrzkt        RealBadAngel, your eb3840a3f88dd9902ff1c5ea50d90a7f33e08d82 didn't work on AMD cards, everything is red
21:03 nrzkt        #3682
21:03 ShadowBot    https://github.com/minetest/minetest/issues/3682 -- tone mapping shader don't compile on AMD cards
21:06 est31        nrzkt, which opengl version?
21:09 celeron55    RealBadAngel develops stuff on nvidia which has the most error-ignoring shader compiler
21:09 nrzkt        mesa versio
21:09 nrzkt        mesa version*
21:09 nrzkt        if i remember it's the last 3.X
21:10 Calinou      radeonsi should support OpenGL 4.1 now
21:10 Calinou      soon 4.2
21:10 nrzkt        Calinou, not atm because some OpenGL 4.0 extensions are not supported
21:10 nrzkt        mesa has 4.1, AMD & intel not
21:11 Calinou      https://mesamatrix.net/
21:11 Calinou      everyone but intel and nv50 do :)
21:11 Calinou      intel lacks two extensions, one in 4.1, one in 40
21:11 Calinou      4.0*
21:11 nrzkt        yes but 4.0 is not enabled for AMD cards in mesa 11.1 atm
21:12 nrzkt        irrlicht used 3.3 then, and if i remember needs 4.2 at least for opengl 4.X
21:15 nrzkt        how can i debug this shader, i didn't have any log :(
21:16 est31        nrzkt, do you use minetest or epixel
21:16 nrzkt        this is the same code for the client
21:16 est31        it has different logging system
21:17 nrzkt        ofc, but all logs are present
21:17 est31        minetest redirects irrlicht errors to log
21:18 est31        when shaders fail to compile, it generates error via irrlicht logging system
21:18 est31        http://irrlicht.sourceforge.net/docu/classirr_1_1video_1_1_i_g_p_u_programming_services.html#a4a8d3b727ee9223d8baa353b82da0478
21:18 LuciousLisa  o/
21:19 LuciousLisa  just here to let you guys know that i just found out about minetest and that you're pretty awesome people.
21:19 est31        I tested it, edited client/shaders/water_surface_shader/opengl_fragment.glsl and added "ddd"
21:19 est31        LuciousLisa, thats great to hear!
21:19 est31        nrzkt, it prints me errors like:
21:19 est31        2016-02-09 22:17:25: ERROR[Main]: Irrlicht: GLSL shader failed to compile
21:19 est31        2016-02-09 22:17:25: ERROR[Main]: Irrlicht: 0:61(1): error: syntax error, unexpected NEW_IDENTIFIER, expecting $end
21:19 est31        2016-02-09 22:17:25: ERROR[Main]: generate_shader(): failed to generate "water_surface_shader", addHighLevelShaderMaterial failed.
21:20 nrzkt        strangely i have irrlicht logs are engine start but not after, looking for the problem
21:20 nrzkt        did you remember where is the irrlicht logging mapping ?
21:21 est31        nrzkt, see commit a93838707a9aab104c20b6a971d8c622184a508f
21:21 nrzkt        yeah irrlicht is mapped too on my server
21:32 nrzkt        est31, in fact the problem was in clientlauncher, mapping was wrong on my log4cpp build :)
21:33 nrzkt        ok i have the error now
21:33 nrzkt        Irrlicht: 0:31(1): preprocessor error: #version must appear on the first line
21:39 est31        yeey my guess turned out to be right :)
21:39 nrzkt        :)
21:39 nrzkt        this permit to fix a bug in my log4cpp interface with irrlicht, that's good :p
21:40 nrzkt        thanks
21:45 nrzkt        this is strange, every shader failed
21:45 nrzkt        selection shader, default_shader, which was added before and worked
21:50 nrzkt        i confirm this is due to the tone mapping thing. I checkout-ed previous commit and everything works properly
21:55 sofar        is there a way to print which shader fails to compile?
21:56 nrzkt        i found the issue
21:56 nrzkt        due to a merge problem, sorry :)
21:56 nrzkt        a little optimisation i didn't reported to MT...
21:56 sofar        yeah but, it's really hard to see which shader fails the issue
21:56 sofar        adding a "glsl compilation failed in shader file "foo"" message may help a ton
21:56 nrzkt        and git add old code i didn't see and make a strange shader :)
22:00 nrzkt        i will report this change to MT in a PR
22:05 nrzkt        est31, can you look at #3683, it's trivial :)
22:05 ShadowBot    https://github.com/minetest/minetest/issues/3683 -- shader.cpp: don't test twice if shader programs are present by nerzhul
22:06 sofar        it's trivial, but does that actually fix something?
22:06 nrzkt        sofar, a PR is not only for fixing things. Here the PR permit to fix the double same test in shader.cpp :)
22:07 sofar        then you should write that in the commit message.
22:07 sofar        :)
22:07 sofar        oh, nvm
22:07 sofar        I misread the title
22:07 nrzkt        write what ? the commit message is explicit :p
22:07 sofar        it's just a double test, nothing really broke
22:08 nrzkt        no, just a little contribution on my spare time i fixed in my MT forked branch that can be backported easily :)
22:15 est31        nrzkt, +1
22:18 nrzkt        if you are okay, then merge when you want :)
22:20 est31        done
22:21 nrzkt        thanks :)
22:24 Fixer        https://github.com/minetest/minetest/issues/329
22:24 Fixer        2 Dec 2012
22:25 Fixer        LOL
22:25 Fixer        and this is not fixed yet
22:28 nrzkt        Fixer, thanks for issue #3680
22:28 ShadowBot    https://github.com/minetest/minetest/issues/3680 -- Merge singleplayer and server tabs (and probably use server method to play local games?)
22:28 Fixer        nrzkt, no problem, feel free to poke it
22:28 nrzkt        i think there is a cleanup to do on singleplayer, not especially server/single tab, but server threading used in single/multi
22:29 Fixer        btw, i'm just now swimming underwater with prelast git on 0.4.12 server and see no drowning, but after taking damage - drowning reappears
22:42 sofar        2016-02-09 14:35:33: ERROR[Main]: generate_shader(): failed to generate "nodes_shader", addHighLevelShaderMaterial failed.
22:42 sofar        what?
22:42 sofar        my minetest is all red
22:42 sofar        I just pulled Tip of Tree
22:42 est31        hmm
22:43 est31        does something come before the code
22:43 est31        some error of some kind?
22:43 sofar        tons of errors
22:43 sofar        pasting
22:43 est31        mhh mine is erronenous too
22:44 est31        nrzkt, you broke it, it seems
22:44 sofar        https://gist.github.com/sofar/e128879eb631b321ef04
22:44 nrzkt        est31 wtf :(
22:44 nrzkt        it fixed for me...
22:44 est31        no
22:45 nrzkt        oh !
22:45 est31        the change is wrong obviously
22:45 nrzkt        there is an inversion in my commit
22:45 nrzkt        if (!vertex_program.empty()) {
22:45 nrzkt        +               pixel_program = shaders_header + pixel_program;
22:45 est31        yeah
22:45 nrzkt        if (!pixel_program.empty()) {
22:45 nrzkt        +               vertex_program = shaders_header + vertex_program;
22:45 nrzkt        those two conditions are inverted
22:47 nrzkt        sofar,
22:47 nrzkt        https://github.com/nerzhul/minetest/commit/e90f7f8585bf55dda4991188c4a6dfb60ff03994
22:47 nrzkt        apply this please and tell me if it's better :)
22:47 sofar        ahhh lol
22:47 sofar        one sec respinning a PR myself atm ;)
22:47 nrzkt        ok, don't do a PR for this i will push it in master if you confirm it's good for you
22:47 sofar        I'm pushing #3608 so it can finally get merged ;)
22:47 ShadowBot    https://github.com/minetest/minetest/issues/3608 -- Fix backface culling when connecting to new servers by sofar
22:48 nrzkt        good sofar, i added a little comment this evening
22:49 est31        nrzkt, tested the patch it fixes the bug
22:49 sofar        I saw, hence the update
22:50 sofar        nrzkt: looks perfectly OK now!
22:50 sofar        after I applied https://github.com/nerzhul/minetest/commit/e90f7f8585bf55dda4991188c4a6dfb60ff03994
22:50 nrzkt        perfect i push it, sorry for the inversion :)
22:51 nrzkt        pushed :)
22:53 sofar        so, now you have time to merge 3608? :)
22:53 nrzkt        not me, but maybe est31 :p
22:56 est31        I'll have a look
23:09 est31        mhh 0.4.13 dev clients before the patch but after commit 735e3b70596e16d04de1edcd878deec3c539c6ed that connect to servers after the patch will probably crash
23:09 est31        but we dont guarantee backwards compat for old dev clients
23:09 est31        so that's okay
23:12 sofar        yup
23:12 sofar        there's not that much in between just yet anyway
23:13 est31        meh, I bet we get bug reports. either way they ought to be closed.
23:16 Fixer        crash? i was having crash today of unknown nature, ntdll one
23:17 est31        sofar, #3503 can now be rebased I think
23:17 ShadowBot    https://github.com/minetest/minetest/issues/3503 -- Nodebox: Allow nodeboxes to "connect" by sofar
23:17 Fixer        could someone swim a little on new clients?
23:18 Fixer        ok, drowning started to work again (randomly)
23:36 Hijiri       is the player pos always the bottom of the collisionbox or is it hardcoded 1m below the collisionbox origin?
23:56 sofar        est31: ty, indeed it needs rebasing I see