Minetest logo

IRC log for #minetest-dev, 2013-11-16

| Channels | #minetest-dev index | Today | | Google Search | Plaintext

All times shown according to UTC.

Time Nick Message
00:03 VanessaE RealBadAngel: the screenshots look nice but they also look like you're shining a spotlight on the landscape
00:03 VanessaE why are the centers of the shots brighter than the sides?
00:04 RealBadAngel intentional
00:04 RealBadAngel specular lighting uses view vector to generate shades
00:05 RealBadAngel it is tuned down, to be barely visible (but noticeable)
00:06 RealBadAngel thx to it surfaces are shadowing themselves
00:06 VanessaE yeah but the screenshots make it look like the light source is right above your head
00:06 VanessaE this is surely not intended.
00:06 VanessaE the light source should be wherever the sun is
00:06 RealBadAngel check it ingame
00:06 VanessaE (and any other light sources, eventually)
00:06 RealBadAngel i mean irl
00:06 RealBadAngel ;)
00:07 RealBadAngel shadows are depending on view angle
00:08 RealBadAngel and i dont mean specular lighting here
00:09 RealBadAngel shadows are done in two ways
00:10 RealBadAngel first is geometry oclussion
00:10 RealBadAngel then bumpmapping
00:10 RealBadAngel angle of camera determines strenght of the effect
00:16 RealBadAngel http://realbadangel.pl/images/tin_ceiling.png
00:16 RealBadAngel example of very detailed, self shadowing texture
00:17 VanessaE code?
00:18 RealBadAngel code is out there for weeks
00:18 VanessaE link?
00:18 RealBadAngel i mean my branch
00:19 RealBadAngel point is to master gimp
00:19 VanessaE you mean the branch with the half-documented shader settings, busted water, parallax mapping that makes textures slide around on the cube surfaces, etc?
00:19 RealBadAngel thats why i have abandoned Haven
00:19 RealBadAngel and created Haven Next Gen
00:20 EvergreenTree joined #minetest-dev
00:20 RealBadAngel water is a subject to another pull
00:20 VanessaE then start using proper branches and forks so this shit can be kept separate!
00:20 RealBadAngel mode 2 of parallax is mixed with mode 1
00:20 RealBadAngel and there are no more modes
00:21 RealBadAngel code limits itself
00:22 VanessaE so, where is the fork+branch of the code where only the solids shaders have been redone?
00:22 RealBadAngel on my hdd
00:22 VanessaE ...
00:23 RealBadAngel i am askin if i can pull it
00:23 VanessaE no one's gonna approve what they can't test reliablt.
00:23 VanessaE reliably*
00:25 RealBadAngel i am playing with evolving HavenNG for weeks already
00:26 RealBadAngel branch is aviable with many builds
00:26 VanessaE the code, damn it.  not the texture files :)
00:26 RealBadAngel jesus christ superstar :P
00:26 RealBadAngel code is weeks old
00:27 RealBadAngel havent changed
00:27 VanessaE yes and is part of a fork that completely breaks the water when shaders are enabled, remember?
00:27 RealBadAngel shit
00:27 RealBadAngel what have i said?
00:28 VanessaE unmolested source or it's all photoshopped :)
00:28 RealBadAngel <RealBadAngel> i would like to push reworked shaders part (solids)
00:28 RealBadAngel <RealBadAngel> without water atm
00:28 VanessaE push != pull
00:29 RealBadAngel semantics
00:29 RealBadAngel u know what i mean
00:29 VanessaE no, I don't.
00:30 RealBadAngel i mean with it: code is stable enough for me to create and enjoy 256x new generation texture pack
00:31 RealBadAngel which leaves all behind
00:31 RealBadAngel and i MEAN it
00:33 VanessaE so post the code to a clean repo somewhere so we can try it out?
00:34 NakedFury joined #minetest-dev
00:39 kahrl there is a feature freeze in effect, this can't go in before 0.4.8 is released
00:40 kahrl (why isn't this in the topic?)
00:47 RealBadAngel kahrl, i dont demand it to be merged prematurely
00:48 RealBadAngel im far from it
00:48 kahrl oh, I misunderstood then
00:48 RealBadAngel im rater askin if i will pull it
00:49 kahrl what do you mean by pull?
00:49 RealBadAngel other devs would like to test it
00:49 kahrl you can always make pull requests, no need to ask
00:49 RealBadAngel because imho its a feature thx to which game can shine
00:50 kahrl quite literally ;)
00:50 RealBadAngel kahrl, to be honest. before i was fighting for each and every line of code to be included
00:51 RealBadAngel now im askin folks to test what have i done
00:51 kahrl there are more core devs now so it might go faster
00:51 RealBadAngel i know the rules and i could use it
00:52 RealBadAngel *them
00:52 RealBadAngel shit,  i just wanted a bit of applause ;) it took me a few months :)
00:53 RealBadAngel and all i got was c55 saying: its not my style
00:54 RealBadAngel kahrl. do you know what attribute for shaders is?
00:55 kahrl I think that's a field in the vertex structure?
00:56 RealBadAngel attribute is an variable set for each shader instance (material)
00:56 RealBadAngel tangent is such a variable
00:57 RealBadAngel there are global variables
00:57 RealBadAngel which we can use
00:57 RealBadAngel but irrlicht is missing per material (per shader instance) variables
00:59 RealBadAngel theres no problem to get normal vector
00:59 RealBadAngel but tangent IS a problem
01:00 RealBadAngel makin the problem more general
01:00 RealBadAngel theres no way to pass variables, vectors, data to shader
01:00 RealBadAngel per instance, not in general, for all of them at once
01:01 kahrl if you only need tangents, you can use S3DVertexTangents I believe
01:01 kahrl but yeah that solution doesn't scale to more attributes
01:01 RealBadAngel i should transform SMesh to IMesh
01:02 kahrl transform?
01:02 RealBadAngel smesh and imesh are not compatible
01:02 kahrl SMesh is a subclass of IMesh
01:03 RealBadAngel from what ive read, to get tangent space in shaders
01:03 RealBadAngel you have to make a copy of mesh
01:03 RealBadAngel with tangents
01:04 RealBadAngel but problem is we are using different ones
01:04 kahrl with IMeshManipulator::recalculateTangents?
01:05 RealBadAngel createMeshWithTangents
01:05 RealBadAngel or so
01:05 RealBadAngel but it would mean to get whole mt world
01:06 RealBadAngel and recalculate tangents for it
01:06 kahrl can't it be done at the end of mesh generation?
01:07 RealBadAngel i do have tangents calculated in mapblockmesh
01:07 RealBadAngel i just need a way to pass it to shader
01:07 RealBadAngel as an attribute
01:09 RealBadAngel i managed to achieve this for a somple switch
01:09 RealBadAngel *simple
01:09 RealBadAngel material texture set to be black or white
01:10 RealBadAngel so i have bound to instance bool value
01:10 RealBadAngel im using it to indicate if normal texture is provided or not
01:11 kahrl isn't it expensive to check that switch in every instance of the shader?
01:11 kahrl wouldn't it be faster to use a different shader depending on the switch?
01:11 RealBadAngel no, its one pixel check
01:12 kahrl still
01:12 RealBadAngel sorry, texel
01:12 RealBadAngel at the start i also thougth so
01:12 RealBadAngel but i do have many effects
01:13 RealBadAngel it ended with cascade of shader instances
01:13 RealBadAngel this off that on
01:13 RealBadAngel endless chain of copy pasta with slight changes
01:14 RealBadAngel now im back to 3 shaders
01:14 kahrl I see
01:14 RealBadAngel lemme point you to the code
01:15 kahrl wasn't there a plan to automatically generate a shader from a set of effects?
01:15 RealBadAngel https://github.com/RealBadAngel/minete​st_GLSLshaders/blob/master/client/shad​ers/solids_shader/opengl_fragment.glsl
01:16 RealBadAngel this doesnt work such way
01:16 RealBadAngel we should render to a buffer for multiple shaders be able to work on the data
01:17 kahrl unfortunately framebuffers can be very slow on less fortunate hardware, or so I heard
01:18 RealBadAngel yes
01:18 RealBadAngel but "if" is not so slow
01:19 RealBadAngel im on 1.8ghz dual core, 4gb ram
01:19 RealBadAngel radeon hd 4670
01:19 kahrl well, I meant more concatenating the scripts for the individual effects to get one shader
01:19 RealBadAngel with 256x textures, with all my changes i do get 60 fps
01:19 kahrl rather than applying each effect as an individual shader
01:20 kahrl that's pretty nice
01:20 RealBadAngel its used box
01:20 kahrl I don't get 60fps with the current shaders :P
01:20 iqualfragile kahrl: please take a look at the pull requests
01:20 iqualfragile there are lots qued up, i think at least some of them deserve merging and all deserve reaction
01:20 RealBadAngel kahrl, my shaders code have some speed ups ;)
01:21 RealBadAngel but still
01:21 RealBadAngel i need some1 to work with me on the subject
01:21 kahrl iqualfragile, there were 80+ a few days ago
01:21 kahrl now there are less than 69
01:21 iqualfragile which is progress
01:21 iqualfragile but not a good state yet
01:22 RealBadAngel ive managed to get oclussion to usable state
01:22 RealBadAngel thx to many weird tricks
01:22 kahrl iqualfragile: I can take a look at them occasionally but don't have time to do it on a regular basis
01:23 kahrl and they just keep pouring in
01:23 iqualfragile i am just worriing a bit…
01:23 RealBadAngel kahrl, would you like to help me with it?
01:23 iqualfragile it seems like nobody would realy feel responsible for minetest development…
01:24 kahrl RealBadAngel: I don't really have the time for it :/
01:29 iqualfragile there are pull requests still open from over a year ago
01:30 iqualfragile maybee create some special date where some coredevs gather?
01:30 iqualfragile merge-monday?
01:31 kahrl well you mean like #238
01:32 kahrl > local fuck = {}
01:32 iqualfragile no, not like that one
01:32 iqualfragile that one would be closable
01:33 iqualfragile more like that one https://github.com/minetest/minetest/pull/237
01:33 iqualfragile it is realy not that much of a change
01:33 iqualfragile and the only thing that people fight about is how long it should sleep
01:34 iqualfragile and then nothing happens
01:34 iqualfragile bikeshed phenomen
01:35 KingsleyT joined #minetest-dev
01:35 kahrl well obviously none of the core devs get 100% cpu usage in the pause menu so it never was an issue for them
01:36 kahrl and finding a value that works for everyone and doesn't look ugly is actually the hard part
01:36 iqualfragile might be true but just setting that to 50 ms would help the case a lot
01:36 iqualfragile you could even make it configurable later on
01:36 iqualfragile but right now nothing changed
01:40 iqualfragile i think the best way to deal with this would in fact be a merge monday where at least 3 coredevs gather and merge or close at least 5 pull requests
01:44 RealBadAngel kahrl, from what i know you were the one that brought shaders to mt
01:44 RealBadAngel so youre the one i should ask for help
01:44 kahrl RealBadAngel: indeed I did
01:45 RealBadAngel i am failing with very basic things
01:45 RealBadAngel passing vectors or tables bound to material
01:46 kahrl although I pretty much only read some tutorial on how to load and compile GLSL/HLSL shaders with irrlicht and made that into a ShaderSource class
01:46 RealBadAngel so did i
01:46 kahrl I don't actually know anything about shader programming
01:46 RealBadAngel last months ive learned a lot about opengl
01:47 RealBadAngel now i can understand what i need
01:47 RealBadAngel and irrlicht engine confuses me
01:47 kahrl iqualfragile: well at the moment only RBA and I are online, and neither of us are very active in the pull requests scene. maybe ask again when other core devs are online?
01:47 kahrl is*
01:47 iqualfragile i did and i got ignored multiple times…
01:48 RealBadAngel did you know that parallax maping is a material type in irrlicht?
01:48 kahrl I guess they don't want to then
01:48 iqualfragile but i have to admid that i was not monitoring minetest actively in the past month or so
01:48 iqualfragile which core devs do you think are most active at the moment?
01:48 RealBadAngel so bumpmapping is
01:49 kahrl RealBadAngel: yeah irrlicht has some odd mix of fixed pipeline and shaders
01:49 RealBadAngel and, to be more as a joke
01:49 RealBadAngel all those effects are hardcoded in asm
01:49 RealBadAngel as a shaders of course
01:49 Warr1024 joined #minetest-dev
01:51 kahrl iqualfragile: can't really say
01:51 iqualfragile is anyone realy active?
01:51 Exio4 they are but at different times
01:51 kahrl the new ones do quite a lot of stuff right now
01:51 iqualfragile names?
01:51 Exio4 shadowninja
01:52 Exio4 and nore? and sapier?
01:52 Exio4 and kahrl is somewhat active, too! ;P
01:52 kahrl BlockMen, nore/Novatux, ShadowNinja, sapier
01:52 Exio4 blockmen is coredev? didn't know
01:53 kahrl heh, I'm just procrastinating in here but not actually working on code
01:59 Topic for #minetest-dev is now Don't forget your scarf, it's feature freezin'. Minetest core development and maintenance. Chit-chat goes to #minetest. Consider this instead of /msg celeron55. http://irc.minetest.ru/minetest-dev/ http://dev.minetest.net/
02:01 iqualfragile oh, wonderfull, now its too late
02:01 iqualfragile i strongly disagree with the timing of this feature freeze
02:02 kahrl the feature freeze has been going on for over a week
02:02 kahrl nobody bothered to put it into /topic for some reason
02:10 iqualfragile oh… ok
02:11 iqualfragile was it announced anywhere?
02:11 kahrl dunno
02:11 iqualfragile will 0.4.8 be the christmas release?
02:12 kahrl yes
02:12 iqualfragile hmm…
02:12 kahrl note that I didn't say christmas 2013 or 2014 or ... ;P
02:12 iqualfragile well, anyways, good night
02:13 kahrl night
02:15 proller joined #minetest-dev
02:17 proller again destroyed env_meta.txt  and all files in players ...
02:38 RealBadAngel kahrl. i will post rework tomorrow,hope u cold review it
03:23 VanessaE ok, we got map corruption going on here along the bottom edge of the world
03:23 VanessaE 22:15:42: ERROR[EmergeThread1]: Invalid block data in database (0,-1933,-1) (SerializationError): deSerializeBulkNodes: decompress resulted in invalid size
03:23 VanessaE third time this has happened, again at the same Y coordinate, but this time at a different X/Z location
03:23 VanessaE 0.4.8 CANNOT GO OUT like this!
03:26 Warr1024 are you using leveldb or sqlite?
03:27 VanessaE sqlite.
03:27 VanessaE I've reproduced this in singleplayer also, on a new map
03:40 kahrl what mods?
03:41 VanessaE well tons of mods of course, but nothing I haven't been using for months on-end
03:41 Warr1024 isn't there some maximum size for a compressed mapblock, so that having too much node metadata, or a very incompressible node arrangement, could cause problems, couldn't it?
03:42 VanessaE the root question though kahrl, isn't what mods do I have (none that can write to the map, afaik), but why the engine is allowing writes to a block that can't possibly exist, and why it's bothering to try to read from that block.
03:43 kahrl well you did mention the bedrock mod
03:43 VanessaE no
03:43 VanessaE I removed that mod, and it turns out that mod wasn't configured to generate a bedrock layer anyway
03:43 VanessaE so about all it did was add a few nodes.
03:44 kahrl because I can't seem to reproduce the same problem with just minetest_game and a new map
03:45 VanessaE I was able to reproduce it simply by teleporting down, clearing a small room, and placing a few torches.
03:45 VanessaE I figure it's ore generation doing it
03:46 VanessaE and when someone did the same on the server, they were able to make it crash by adding a travelnet box, which is just a normal node that triggers the teleport-player function
03:46 VanessaE (with a formspec)
03:48 kahrl maybe I need to try a different seed
03:48 kahrl restarted the server a couple times, dug a lot of nodes, nothing so far
03:49 VanessaE one sec and I'll upload the current iteration of my game
03:50 kahrl oh!
03:50 VanessaE http://digitalaudioconcepts.com/vanessa​/hobbies/minetest/VanessaE_Game.tar.bz2
03:50 kahrl I managed to get the "suspiciously large amount of objects detected" message
03:50 VanessaE oh good!
03:50 VanessaE that's what I was getting in singleplayer
03:50 kahrl by punching a pile of gravel that generated at the very bottom
03:51 VanessaE now add a travelnet box ( https://forum.minetest.net/viewtopic.php?id=4877 )  there and at the surface, and transport up/down a few times
04:03 kahrl the longer you stay at the bottom of the map at a time, the larger the number of removed objects gets
04:04 kahrl I suppose this makes the mapblock hit some kind of sqlite blob size limit or something
04:05 Warr1024 maybe the mapgen should just stop generating falling_node stuff below a certain depth?
04:05 Warr1024 or alternative, deal with them in some better way if they fall into oblivion
04:07 kahrl I haven't waited long enough to actually hit the limit yet (it hasn't crashed yet)
04:07 kahrl but yes something needs to be addressed
04:07 kahrl I wonder what it does if I were to apply #1000
04:07 VanessaE thing is, in my singleplayer test, there were NO entities in the area.
04:08 VanessaE the number of objects must be the result of misreading some other field
04:08 VanessaE kahrl: I have #1000 applied already
04:10 kahrl how long did you have to stay at the bottom for it to crash?
04:10 VanessaE idk, I didn't get a crash out of the singleplayer test, but cheapie managed to get several crashes in a row within a couple of minutes after it started
04:11 VanessaE all I got out of singleplayer was the same error you got
04:13 kahrl I didn't get the message until I played with entities (made the gravel fall or dropped some item into oblivion)
04:14 VanessaE maybe because I have wield3d
04:14 VanessaE it uses an entity attached to the player's hand to create the wielded item
04:14 kahrl ah, yeah
04:16 kahrl 20000000 objects (which took around 5 minutes to accumulate) are not enough to trigger the crash
04:16 VanessaE it may not crash in single player at all
04:16 VanessaE try running that map in a standalone server maybe?
04:17 kahrl unlikely since singleplayer is a server and a client
04:18 kahrl I just suspect the guy on your server was doing something else and had the offending mapblock loaded for a long time
04:18 VanessaE I'd have thought so too, but I've had instances in the past where a singleplayer combo behaved very differently from separate server/client instances
04:19 kahrl I might go to sleep and leave minetest running
04:21 kahrl maybe not... it's starting to swap
04:21 VanessaE try it as separate instances first.  might save some time
04:21 VanessaE oh my!
04:22 RealBadAngel kahrl, will you be there tommorow at the same time?
04:23 kahrl I don't know
04:23 RealBadAngel what is your time zone btw?
04:24 kahrl cat /etc/timezone => Europe/Berlin
04:24 RealBadAngel shit!
04:24 RealBadAngel i was living in Berlin for a few yrs
04:25 RealBadAngel Ick bin ein Berliner ;)
04:26 kahrl :)
04:27 hmmmm ryan@titan:/home/ryan % cat /etc/timezone
04:27 hmmmm cat: /etc/timezone: No such file or directory
04:27 hmmmm what
04:27 VanessaE kahrl: the reason it actually crashes for me might also be because I have like, a metric assload of RAM.  maybe it reaches some "totally b0rk3d" point before I run out of RAM?
04:27 RealBadAngel no clocks on titan?
04:28 kahrl VanessaE: sounds plausible
04:29 VanessaE I had Survival suddenly go off into la-la land earlier.  It hit 11GB before the OOM killer nailed it
04:29 RealBadAngel btw, VanessaE complained before bout 512x not workin anymore on her linux
04:29 hmmmm sapier didn't commit his RequestQueue patches?
04:29 VanessaE (though Creative is where the map bug happened)
04:30 RealBadAngel i were using 128x on debian and got 30-40fps
04:30 VanessaE 512px works, RBA, but unusably slow at the spawn on either of the two above servers.  Different subject entirely.
04:30 RealBadAngel now im using enhanced shaders 256x, fresh win7 instal
04:31 RealBadAngel 60+ fps always
04:31 VanessaE hmmmm: I have that patch on my client, though idk if it's up-to-date.
04:32 VanessaE RealBadAngel: but is it in the same map and area of the map?
04:32 RealBadAngel even when loggin onto your server
04:32 VanessaE at the creative spawn?
04:33 hmmmm anyway I suppose that a big part of the reason why the original emergethread was so unstable was because it used RequestQueue
04:33 hmmmm which corrupts memory
04:33 VanessaE RealBadAngel: go ahead.  I dare you :)
04:33 RealBadAngel what could i say
04:34 RealBadAngel Thorvalds already shown them a finger
04:35 RealBadAngel its all about drivers
04:35 RealBadAngel mt is way faster on windiows
04:36 VanessaE of course on my system, I've noticed things like windows that don't properly save and restore their contents, so a terminal under a minetest window will make the minetest window flicker and jump as it scrolls
04:36 RealBadAngel at least client
04:37 RealBadAngel 30fps max on lnux and 60+fps on windows
04:38 RealBadAngel sad to say, but windowss is cool to test my shaders code
04:39 VanessaE so, is minetest/irrlicht just not using the linux drivers to their fullest?  surely there's a reason
04:42 VanessaE maybe c55 is right and we need to bring the texture atlas back...but make it way bigger
04:43 VanessaE (darkrose made minetest-classic's atlas 4x the size of 0.3.3's.  512px works great there)
04:47 VanessaE kahrl: so...anything interesting yet? :)
05:09 werwerwer joined #minetest-dev
05:13 Warr1024 so this whole thing runs on udp, so it has to do retransmissions manually, right?
05:13 Warr1024 does anyone know how long it takes for the server to retransmit during a player emerge?
05:14 Warr1024 I keep getting stuck at downloading Node Definitions, probably due to a lost packet due to wifi timing shit.
05:15 djdduty joined #minetest-dev
05:24 Zeitgeist_ joined #minetest-dev
05:24 Zeitgeist_ joined #minetest-dev
05:31 hmmmm there's TCP but apparently it doesn't work right for some reason and nobody can really investigate because nobody has the time to do so
05:32 hmmmm on another note, it's amazing what funding can do.  I can really understand now why celeron wanted to try to monetize minetest or at least a derivative of it
05:33 hmmmm if people had a way to live off of working on minetest they'd sure as hell find time to work on this stuff
05:35 Warr1024 it would probably be pretty hard to make enough to support more than one developer...
05:35 Warr1024 so at best, you might be able to swing one full-time guy.
05:35 Warr1024 seems like you'd be lacking something though without other members of the core team...
05:35 hmmmm that's definitely true, everybody brings something to the table
05:36 Warr1024 yeah, and it's not like you can spread the funding around so that each of the core devs can quit like 2% of his/her job.
05:37 Warr1024 of course, a full-time reviewer and integrator (with some time left over for some dev) might not be bad.
05:46 Exio4 joined #minetest-dev
05:55 OldCoder joined #minetest-dev
05:57 RealBadAngel =====================================​=====================================​==================================]]
05:58 VanessaE allright who let RBA's cat onto thekeyboard?
06:03 OldCoder Can he or she code?
06:19 OldCoder joined #minetest-dev
06:20 Zeitgeist_ joined #minetest-dev
07:09 Akien joined #minetest-dev
08:02 darkrose joined #minetest-dev
08:02 darkrose joined #minetest-dev
08:17 Akien joined #minetest-dev
08:42 deltib joined #minetest-dev
08:46 Calinou joined #minetest-dev
09:38 Jordach joined #minetest-dev
10:13 Jordach joined #minetest-dev
10:13 Akien joined #minetest-dev
10:22 Taoki[mobile] joined #minetest-dev
10:37 PilzAdam joined #minetest-dev
10:48 EdB joined #minetest-dev
11:11 iqualfragile joined #minetest-dev
11:31 RealBadAngel joined #minetest-dev
11:44 werwerwer joined #minetest-dev
12:13 nore joined #minetest-dev
12:27 nore does anyone know why proller closed all his pull requests?
12:28 ImQ009 joined #minetest-dev
12:34 nore celeron55, is #1000 finished, or should we wait a bit more for it?
12:39 celeron55 has VanessaE reported that it works?
12:39 celeron55 or someone else with some heavy testing on it
12:57 Jordach celeron55, i've seen client side dupes, but no actual true dupes
12:59 Jordach celeron55, http://i.imgur.com/zrlAcfz.png
13:00 Jordach 1000 works, but client side dupes happen too
13:14 proller joined #minetest-dev
13:14 troller joined #minetest-dev
13:16 iqualfragile proller: why did you close all your pull requests?
13:19 EvergreenTree joined #minetest-dev
13:30 troller iqualfragile, they joined into one: https://github.com/minetest/minetest/pull/1010
13:31 iqualfragile troller/proller that is quite a bad idea
13:31 troller i cant support lot of pulls, too many conflicts
13:31 iqualfragile the chance of getting that merged are minimal
13:31 troller its not my problem
13:33 troller i want to make something good, and i cant with 20+ pulls
13:36 NakedFury joined #minetest-dev
13:42 hmmmm joined #minetest-dev
13:44 PilzAdam lol, troller, that will never be merged
13:44 troller as other pulls..
13:48 iqualfragile PilzAdam: what would you think about a merge-monday
13:48 PilzAdam I dont have time at mondays
13:49 iqualfragile where you (as in the core devs) force yourselve(s) to merge or close/comment at least 5 pull requests
13:49 iqualfragile well, it does not have to be a monday
13:49 iqualfragile because right now the situation is horrible
13:49 iqualfragile there are lots of people willing to contribute to minetest but they get frustrated because of missing feedback
13:49 PilzAdam we actually merged quite a lot of pull requests last weekend
13:50 iqualfragile i know, around 12
13:50 iqualfragile which is progress
13:50 iqualfragile but thera are still tons of pull requests
13:50 zat joined #minetest-dev
13:50 iqualfragile and some are more then a year old
13:50 PilzAdam the problem with a set task is that the quality might suffer
13:50 iqualfragile troller/proller i would recommend to close this pull request and reopen the smaller ones
13:51 iqualfragile this one will never be merged
13:51 iqualfragile PilzAdam: how so?
13:51 iqualfragile as i said 5 to either merge or close/comment
13:51 PilzAdam if we merge for the sake of merging and not to get good code then the quality may decrease
13:53 iqualfragile PilzAdam: as i said: merge or close/comment
13:53 iqualfragile the only thing it would cost would be a bit of time
13:57 iqualfragile additionaly there are quite some pull requests which are quite good but need some kind of litle modification to be perfect, those modifications would cost somebody who knows his way around in the minetest code 20 minutes max
13:59 nore I have a question: should I rewrite completely force loading? If so, could you tell me what you would think of the API I'm thinking of?
14:01 nore the idea: minetest.get_forceload_handler() -> returns a force_load handler
14:01 nore handler has a few methods: -> force_load_block(blockpos)
14:01 iqualfragile for example 385 or 237
14:02 nore returns true or false if it can be force_loaded
14:02 nore -> free_block(blockpos) --> idem
14:02 nore and perhaps some get_loaded_blocks too
14:03 nore but the idea would be a limit of forceloaded blocks for each mod
14:03 nore and no mod can force load more than that limit of blocks at a time
14:03 nore so, what do you think of that?
14:08 iqualfragile joined #minetest-dev
14:13 iqualfragile nore: force loading is for letting blocks stay active even when all the players leave the area, right?
14:13 nore yep
14:13 iqualfragile so for example for farming
14:13 iqualfragile i think the beter way to deal with this would be to save some timestamp of when the block was unloaded
14:13 iqualfragile and then pass the difference between the current time and the timestamp to every node in the block so they can react acordingly
14:13 nore that's already done
14:13 iqualfragile it is?
14:14 nore (there is a pull for that for furnaces working when unloaded wy simulating them again)
14:14 nore however, the objective is to be able to change the area even when players are far away
14:15 nore for example, there could be a technic quarry that mines items, and sends them 1km further, where a player gets it
14:15 PilzAdam https://github.com/minetest/minetest/pull/934 any objectios?
14:15 nore to access the area too, and to be able to keep entities active mainly
14:16 nore PilzAdam, none
14:16 iqualfragile nore: interesting idea
14:16 iqualfragile and in fact a totaly different usecase
14:16 iqualfragile can you point me to that other pull request?
14:16 nore you mean, the current one?
14:17 iqualfragile PilzAdam: can you explain why there are mapgen_x materials?
14:17 nore https://github.com/minetest/minetest/pull/606
14:17 PilzAdam iqualfragile, the mapgenerator outputs mapgen_* to not depend on certain mod prefixes
14:18 PilzAdam its up to the game to decided which material to place at the position of e.g. dirt
14:18 iqualfragile nore: nah, i meant the one where the passed time is passed to the nodes upon reloading the block
14:18 nore ah, it is https://github.com/minetest/minetest_game/pull/200 IIRC
14:18 PilzAdam nore, should I add a mapgen_ice alias to minetest_game?
14:19 nore PilzAdam, I think so (dunno if 0gb_us has already done that though)
14:19 PilzAdam there is no pull request from him
14:19 nore then yes, you should add an alias for that...
14:19 iqualfragile nore: well… that one is just for furnaces…
14:20 iqualfragile it would be nice to have some kind of uniform thing for that
14:21 PilzAdam hmmmm, what do you think about https://github.com/minetest/minetest/pull/934 ?
14:21 nore iqualfragile, yes, I know... perhaps an ABM parameter max_block_reload_simulate_times or something like that...
14:21 hmmmm ugh
14:21 hmmmm I'll scoop it up with my latest mapgen v7 revamp
14:22 hmmmm I feel like my mountains are starting to look better now that I'm using an exponential height decay
14:22 PilzAdam so #934 can be closed?
14:22 iqualfragile PilzAdam: thanks for working on pull requests :D
14:23 hmmmm it can
14:24 PilzAdam iqualfragile, Im quite limited since we have a feature freeze...
14:25 iqualfragile PilzAdam: yes i know
14:25 iqualfragile i think the timing for this feature freze is quite bad
14:25 nore PilzAdam, what would you think of that new API for force_loading? should it even be done?
14:26 iqualfragile there are some pull request which would be wonderfull to have in the christmas release
14:26 iqualfragile will list them
14:28 PilzAdam when you look at the pull request list then youll notice that pages >= 2 only contain feature enhancements
14:29 nore PilzAdam, ^
14:32 hmmmm hmmm
14:32 PilzAdam https://github.com/minetest/minetest/pull/1006 I can reproduce the crash and can confirm that it fixes it
14:32 PilzAdam it was probably copied code from "tiles", which allowes 6 entries
14:35 PilzAdam any objections to merging it?
14:35 iqualfragile joined #minetest-dev
14:35 nore nope
14:35 hmmmm sounds good
14:36 hmmmm hmmm
14:37 Taoki joined #minetest-dev
14:38 nore why does nobody want to tell me anything about the force loading new API proposal?
14:38 PilzAdam nore, maybe people need to think about it first?
14:39 nore yeah, but I had not even a word... but I can wait a bit... ;)
14:40 iqualfragile ok, categorized them: first fixes ordered in decreasing order of importance: https://github.com/minetest/minetest/pull/1000 https://github.com/minetest/minetest/pull/744 https://github.com/minetest/minetest/pull/977
14:42 iqualfragile next: graphical goodies again most important first: https://github.com/minetest/minetest/pull/799 https://github.com/minetest/minetest/pull/385 https://github.com/minetest/minetest/pull/583
14:48 PilzAdam this seems good: https://github.com/minetest/minetest/pull/998 any comments?
14:48 nore yep: it is good
14:50 iqualfragile joined #minetest-dev
14:50 hmmmm where is that sapierrrr
14:50 iqualfragile joined #minetest-dev
14:51 hmmmm do you think we should just commit his requestqueue patches for him and release?
14:51 iqualfragile nearly done: next gameplay improvements: https://github.com/minetest/minetest/pull/816 (this one would be especialy awsome because minecraft does not have it :D) https://github.com/minetest/minetest/pull/866 https://github.com/minetest/minetest/pull/958
14:51 PilzAdam we need to fix https://github.com/minetest/minetes​t/issues?milestone=1&amp;state=open too
14:51 iqualfragile and then there is this one: https://github.com/minetest/minetest/pull/1001 which would be quite important for server administrators fighting against griefers
14:51 iqualfragile if these could get a secound look i would be happy
14:52 iqualfragile nearly done: next gameplay improvements: https://github.com/minetest/minetest/pull/816 (this one would be especialy awsome because minecraft does not have it :D) https://github.com/minetest/minetest/pull/866 https://github.com/minetest/minetest/pull/958
14:52 iqualfragile looks like quassele would not handle connection losses very well
14:55 PilzAdam nore, I merge it then
14:55 nore ok (but I didn't test it though)
14:56 PilzAdam well, I have tested it (and I guess sapier too)
14:59 PilzAdam iqualfragile, I merged or commented on nearly all pull requests that are mergeable during a feature freeze now
15:12 iqualfragile ok, thanks
15:13 hmmmm minetest.get_connected_players().... wasn't that optimized by that one guy?
15:13 hmmmm just recently
15:13 hmmmm that could probably be the source of the problem
15:14 PilzAdam https://github.com/minetest/minetest/commi​t/2170c9958a455c228459c795857ec1a890cd05c5
15:14 hmmmm i suppose what could be done is revert that patch and figure it out for after 0.4.8
15:14 PilzAdam that is already linked in the issue
15:14 hmmmm do you see any obvious causes for the bug in that code?  i'm looking at it right now
15:15 PilzAdam I already wrote about it in the issue
15:15 PilzAdam we could check value:is_player() when creating temp_table in misc.lua
15:16 hmmmm hmmm the code looks solid
15:17 hmmmm see, I'm juggling around 3 scripting languages every day, so forgive me for asking this stupid Lua question:
15:17 hmmmm if table.foo is never set to anything, it's nil, right?
15:17 nore yep
15:17 PilzAdam yep
15:17 hmmmm for k, v in pairs(table) would only retrieve values that are not nil
15:17 hmmmm regardless of whether they've been set to nil before or not
15:18 nore yep
15:18 nore wait... so why does the code crash?
15:18 hmmmm so that can't be the source of the problem
15:21 hmmmm oh disregard that, I didn't realize you already found what's wrong
15:21 nore PilzAdam, since you merged #998, you should close #996 and #997
15:21 hmmmm yeah you should add that is_player check
15:21 nore ah, p being nil... (I thought it was player too)
15:22 PilzAdam nore, thanks for reminding me
15:22 nore (I'm quite surprised #997 got labeled as blocker, though)
15:24 Zeitgeist_ joined #minetest-dev
15:32 PilzAdam VanessaE, can you test if the crash reproduces with this patch: https://gist.github.com/PilzAdam/7501466 ?
15:32 PilzAdam (the one in simple mobs has to be removed)
15:35 hmmmm PilzAdam, when is that code you pasted in the issue executed?
15:35 hmmmm on step or something?
15:35 PilzAdam on_step of each mob
15:35 hmmmm okay
15:36 hmmmm well Environment::step is run after handlePeerChanges()
15:37 hmmmm so handlePeerChanges runs, calls on_leaveplayer, which updates the external lua player list
15:38 hmmmm so lemme check if objects are updated between the time those two execute
15:39 hmmmm which they aren't
15:40 hmmmm anyway, the "official" way of getting around this problem is to ignore players with a peer_id of 0.  can you access that from lua?
15:42 PilzAdam AFAIK no
15:43 hmmmm I say that checking is_player() won't work since that only checks to see if the reference to the player is not null; which it won't be since the object list hasn't yet been updated by the time the environment step runs
15:43 hmmmm so the recommended fix for this, I'd say, is to introduce a new API that gets the peer id so you can compare against that.
15:44 PilzAdam is it for any use for mods? or do we just offer it to builtin?
15:44 hmmmm does it matter?
15:44 PilzAdam adding useless stuff to the API is not really good
15:46 hmmmm well, alright.  since nobody mentioned wanting a peer id from lua, I suppose it could just be a 'middle' function for builtin but not for the end user
15:46 hmmmm call it, ObjectRef::is_player_connected()
15:46 hmmmm l_is_player_connected() rather
15:47 hmmmm ObjectRef *ref = checkobject(L, 1); Player *player = getplayer(ref); lua_pushboolean(L, player && player->id != 0); return 1;
15:48 hmmmm make that player->peer_id
16:06 Calinou joined #minetest-dev
16:24 rubenwardy joined #minetest-dev
16:28 OldCoder joined #minetest-dev
16:51 zat joined #minetest-dev
16:53 celeron55 proller: what the fuck man
16:54 proller what?
16:54 celeron55 proller: that's right up being a prick and will only cause harm to everyone; it won't be merged and probably the stuff that it tries to fix will be reverted to their original state...
16:55 celeron55 your pull won't and can't be merged; there's no way it will happen because nobody can review that amount of stuff
16:55 hmmmm minetest-proller
16:55 hmmmm I would actually say that there's enough material there to make your own fork
16:56 proller lot of my pulls also "won't and can't be merged" 3+months
16:56 proller there is only one problem, how to name fork
16:57 hmmmm well, it heavily changes the game dynamics with all the weather and finite liquid related stuff
16:58 PilzAdam proller, my impression is that you dont want to create a nice game engine but a nice game; maybe forking 0.3 and bringing all your weather and finite liquids stuff in would suit you better?
16:58 hmmmm indeed, I think pilzadam is 100% right on that
17:00 proller 0.4 is good enough
17:01 rubenwardy_ joined #minetest-dev
17:02 celeron55 it would be good if ways to make weather possible without raping the engine nor raping the contributor who is making it were found
17:03 celeron55 but if not, then... i think the right action is to not add it and let the possibility emerge from more primitive things that are going to be added in the future
17:14 celeron55 the only strong combination of agruments for adding weather in the "proller way" requires the extra supposition that development is going to slow down in the future so much that it would never get done otherwise
17:15 celeron55 that's hard to predict, but predicting stuff in a short-sighted way like that can be a self-fulfilling prophecy 8)
17:17 celeron55 by the way, does anyone know how to reproduce client-side dupes as mentioned by jordach?
18:18 zat joined #minetest-dev
18:23 djdduty joined #minetest-dev
18:55 artur99 joined #minetest-dev
18:56 artur99 hi
19:07 OldCoder Hi
19:08 OldCoder artur99, you had a font question?
19:08 OldCoder https://github.com/minetest/minete​st/blob/master/src/main.cpp#L1411
19:55 artur99 Romanian translatio is DONE! :D
19:55 artur99 *translation
20:16 OldCoder Ah
20:16 OldCoder You did announce it. PilzAdam ^
20:39 OldCoder Hi. Reporting a possible bug that has now been observed by multiple people. Sometimes game load stops on Item Definitions and never starts again. Restarting world fixes it. Is this a possible network issue or other problem?
20:39 OldCoder Issue has been observed in different worlds by different people. Server is git about a week or two old.
21:04 IceCraft joined #minetest-dev
21:13 rubenwardy_ joined #minetest-dev
21:46 thexyz OldCoder: how long did you wait before trying to reconnect again?
21:46 thexyz OldCoder: I think I had a related issue where server reports that it cannot emerge player
21:46 thexyz but after 1-2 mins I can connect to it
21:48 OldCoder thexyz, They were stuck for half an hour
21:48 OldCoder Multiple people
21:48 OldCoder It is happening to me right now
21:49 OldCoder Noticed it a week ago but waited to test before reporting it
21:49 thexyz what does the server output?
21:49 OldCoder Output where? In debug?
21:49 OldCoder No special messages in debug.txt. Item definitions simply hang.
21:49 OldCoder Happening right now to me
21:50 OldCoder harrison had it earlier but I restarted the world and then he got in
21:52 OldCoder thexyz, I could restart the world and get in but that would eject harrison. Suggestions are welcome.
21:52 thexyz no idea, sorry
21:52 OldCoder k
21:52 OldCoder thexyz, I think this is a core bug. I will repeat every 2 days or so. Is that all right?
21:52 OldCoder Waited a week before speaking. And it is a show stopper.
21:52 thexyz I think you should create an issue
21:52 OldCoder How is that done?
21:53 OldCoder And should I try rebuilding with git current first? But my copy is only about a week old.
21:53 OldCoder I restarted client twice and then could get in
21:53 thexyz https://github.com/minetest/minetest/issues/new
21:54 OldCoder thexyz, thank you!
21:54 OldCoder thexyz, I shall rebuild and test with latest. Then report.
22:38 Taoki joined #minetest-dev
22:55 EvergreenTree joined #minetest-dev
22:59 KingsleyT joined #minetest-dev
23:05 iqualfragile joined #minetest-dev
23:10 iqualfragile what is the general plan for minetest 0.4.8? for example: should mgv7 be in it?
23:49 iqualfragile what license does the minetest icon use?
23:49 PilzAdam CC BY-SA 3.0
23:54 iqualfragile PilzAdam: where can i find that?
23:54 iqualfragile not the license, the notice that the icon uses that license
23:55 PilzAdam https://github.com/minetest/mine​test/blob/master/README.txt#L253
23:56 salamanderrake joined #minetest-dev

| Channels | #minetest-dev index | Today | | Google Search | Plaintext