Minetest logo

IRC log for #minetest-dev, 2014-01-02

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

All times shown according to UTC.

Time Nick Message
00:00 Weedy joined #minetest-dev
00:09 sapier https://github.com/minetest/minetest/pull/1078 TCP/UDP mixed mode support please review until I have enet variant ready for testing
00:09 sapier btw I still don't really know how to compare those variants any ideas are welcome
00:31 us_0gb joined #minetest-dev
00:53 OldCoder joined #minetest-dev
02:31 Miner_48er joined #minetest-dev
02:49 sapier left #minetest-dev
03:51 djdduty joined #minetest-dev
04:04 bas080 joined #minetest-dev
04:23 salamanderrake joined #minetest-dev
05:13 OldCoder joined #minetest-dev
05:13 OldCoder joined #minetest-dev
06:43 us_0gb joined #minetest-dev
06:51 xyz cool in the end hmmmm just fucking disappeared
06:57 xyz well I guess no one else is interested
07:01 kahrl guess that means more time to write the changelog
07:02 kahrl ah, you already did that
07:04 hmmmm ???
07:04 hmmmm I am here
07:04 hmmmm I thought the release was done
07:07 RealBadAngel hi  folks
07:07 kahrl did the clientmedia fix for webservers that can't deal with POST go in?
07:08 kahrl hrm, apparently not. oh well
07:08 kahrl I kind of forgot about that one
07:08 RealBadAngel i was readin lately the chan
07:09 RealBadAngel do we have a deal who is responsible for specific parts of code?
07:10 kahrl RealBadAngel: yeah, subsystem maintainers similar to how the linux kernel is maintained
07:10 RealBadAngel because i would like to sign specific seat
07:11 RealBadAngel thats in general good idea imho
07:11 RealBadAngel since we accomodated a few coders, workin in specific areas
07:12 kahrl right
07:13 RealBadAngel i would like to be responsible for shaders
07:13 kahrl here is the current list of subsystems: https://gist.github.com/celeron55/8189279
07:13 kahrl shaders would be subsumed under client/audiovisuals
07:14 RealBadAngel i can see
07:14 RealBadAngel so i should be with c55 in a team?
07:14 us_0gb Testing ...
07:14 RealBadAngel disease
07:15 kahrl no, the idea was to have only one maintainer per subsystem
07:15 RealBadAngel minimalistic theory vs eyecandies?
07:15 RealBadAngel thats not gonna work ;)
07:15 kahrl but c55 wanted to give the subsystem to someone else soonish, you should talk to him
07:15 RealBadAngel not in audiovisuals
07:16 RealBadAngel c55 way was pc at (<-ideal)
07:17 RealBadAngel my way is as much eyecandies as it is possible
07:17 RealBadAngel and targeting new GPU's
07:19 RealBadAngel btw im suprised he had chosen that area
07:19 kahrl http://irc.minetest.ru/minetest-dev/2013-12-30#i_3522914
07:23 RealBadAngel controversial
07:24 RealBadAngel i was learning GLSL for months
07:24 RealBadAngel i made working shaders
07:24 kahrl he means from the artistic point of view
07:24 kahrl the code is fine
07:25 RealBadAngel because im the only one texture pack maker that made a pack?
07:25 RealBadAngel fuck i even added auto generation
07:25 RealBadAngel i hate c55 attidude
07:25 nore joined #minetest-dev
07:26 RealBadAngel it is not acceptable
07:26 RealBadAngel i wont be crawling on the floor for ages
07:26 kahrl as I have zero artistic clue, you'll have to work this out with c55 yourself :/
07:27 RealBadAngel i do have next part of code, tested for months
07:27 RealBadAngel ready to merge
07:27 RealBadAngel but i wont pull it
07:28 RealBadAngel that what i have read was enough for me
07:28 RealBadAngel please sign me outta this organisation
07:28 RealBadAngel im going back to modding only
07:28 kahrl ...
07:29 RealBadAngel im done with core development
07:29 kahrl I see no problem with adding "controversial" shaders as long as they can be disabled
07:29 RealBadAngel say it to c55
07:29 RealBadAngel im off that business
07:30 RealBadAngel left #minetest-dev
07:34 nore so, 0.4.9 was released?
07:40 RealBadAngel joined #minetest-dev
07:40 RealBadAngel back for just one thing
07:41 RealBadAngel c55 taking care of audiovisuals. things that require new hardware and power. what a joke.
07:41 RealBadAngel ^for the record
07:42 RealBadAngel left #minetest-dev
07:47 xyz eh
07:48 xyz overreacting over those simple things is no good
07:48 blaaaaargh joined #minetest-dev
07:49 hmmmm erm
07:50 hmmmm I don't know if I like that division of subsystems and having "exactly" one person for each
07:50 hmmmm for example you can't put sapier on the server and client, many people have and do modify it
07:51 hmmmm I can't put sapier on any single one subsystem
07:51 hmmmm and the scriptapi, everybody modifies that when they add new api of their own
07:51 hmmmm nobody should be able to (or should I say can) own scriptapi
07:51 xyz okay, I've published github release and posted forum entry
07:52 hmmmm not to mention that this adds a dependency on talking to these specific people in order to modify something if this plan is to really come to fruition
07:52 hmmmm the bus factor becomes an issue
07:53 Naked joined #minetest-dev
07:54 xyz and the website is updated too
07:55 xyz Zeitgeist_: please update gentoo ebuild for Minetest, 0.4.9 is released now
08:00 Zeitgeist_ xyz: kk gonna update the ebuild s fast as possible
08:00 Zeitgeist_ as*
08:13 darkrose joined #minetest-dev
08:45 blaaaaargh joined #minetest-dev
10:15 john_minetest joined #minetest-dev
10:21 celeron55 eh what
10:22 celeron55 it's not about who can modify something; it's about who is responsible for making sure pull requests get handled and fixing the sybsystem if it doesn't work
10:24 celeron55 wtf is this thing with rba
10:26 celeron55 he just suddenly pissed himself off
10:28 celeron55 apparently he thinks i am going to suddenly stop anything made by him getting into upstream or something
10:28 celeron55 while i wasn't even around to be commenting anything
10:28 celeron55 that's fair
10:33 celeron55 also i already previously said that the "allocated" persons can simply give the same permission to someone else too if they see fit; but it's up to the one persons to handle that, not me or someone else
10:34 celeron55 apparently people are too used to the current organization to be able to think individually like that
10:34 celeron55 john_minetest: no, but there should be
10:35 celeron55 it should be a page which also overally describes the idea
10:35 celeron55 "Organization" or something
10:36 celeron55 i think dev wiki is best
10:38 BlockMen joined #minetest-dev
10:41 BlockMen is there a special reason that 0.4.9 is released already?
10:42 nore BlockMen, yes: we got back to normal release time... not that very long time before 0.4.8
10:44 BlockMen ic
10:48 celeron55 https://gist.github.com/celeron55/8189279
11:06 RealBadAngel joined #minetest-dev
11:10 sfan5 what the heck? ndk debugging is broken on 4.3... now I need to buy another device to debug minetest or what? -.-
11:11 sfan5 ( https://code.google.com/p/android/issues/detail?id=58373 )
11:11 Calinou joined #minetest-dev
11:19 RealBadAngel celeron55, sorry, i just got upset because you were always the one blocking most of eyecandies i was working on
11:21 Robby joined #minetest-dev
11:28 eeew joined #minetest-dev
11:39 SmugLeaf joined #minetest-dev
11:39 SmugLeaf joined #minetest-dev
11:39 celeron55 i updated the text on http://dev.minetest.net/Organisation
11:40 celeron55 if anything is unclear to anyone, ask me and i will answer on that page
11:41 celeron55 i guess we should define how the maintainers are selected and when they change
11:43 celeron55 or i should
11:44 RealBadAngel celeron55, so... https://github.com/RealBadAngel/minetest/commit/a48d0e9bed5d478e1b9c468484eb81420ffed28b
11:45 RealBadAngel this is waiting for review i think...
11:51 celeron55 does someone use this?
11:52 RealBadAngel i was asking many times for testing...
11:52 RealBadAngel this change alone makes this: http://i.imgur.com/Jh13V2j.png
11:53 RealBadAngel without any extra files needed
11:54 RealBadAngel it was intended for 16x textues, but works suprisngly well for any
11:56 RealBadAngel it uses Sobel's 3x3 operator for normals and for each texel 3x3 gausian blur
11:58 RealBadAngel 256x sphax: withou: http://i.imgur.com/LEzebiX.png and with: http://i.imgur.com/RX2BUjh.png
11:58 pitriss RealBadAngel: maybe you should fix your shaders to not break look of framed glass.. With shaders are large glass places just ugly.. It is weird to see sides of nodes inside glass..
12:00 RealBadAngel i havent got such issue report
12:02 kaeza joined #minetest-dev
12:03 RealBadAngel joined #minetest-dev
12:04 RealBadAngel sorry. power was down here for a short while
12:05 RealBadAngel pitriss, please open an issue with screenshots how it looks like for you, your os, GPU used, drivers, and log for this session
12:05 jin_xi joined #minetest-dev
12:06 pitriss ok i will do that later..
12:07 RealBadAngel GLSL is very hard to maintain since same code works different for each version and if that was not enough for different GPUs (ie AMD or NVIDIA)
12:10 RealBadAngel without feedback i cant do nothing, because i do own only AMD GPU
12:10 Jordach joined #minetest-dev
12:11 celeron55 http://dev.minetest.net/Organisation#proposed:_releases
12:11 celeron55 i (and many others) would be glad if we could get someone on that
12:11 ImQ009 joined #minetest-dev
12:12 RealBadAngel celeron55, any comments on screenshots?
12:28 celeron55 all i can see is that this patch doesn't do anything for me
12:28 celeron55 ah, now it does
12:29 celeron55 lol, this looks ridiculous
12:30 celeron55 nowadays i can test on both, a haswell and an nvidia, on thee same computer 8)
12:30 celeron55 it slows down the haswell so much that it's unplayable
12:35 Jordach celeron55, i'll give it a test - see how bad it is
12:35 iqualfragile joined #minetest-dev
12:38 celeron55 anyway i can merge this
12:38 celeron55 i guess if we don't have a good game, then we have to let people play with silly shaders 8-)
12:39 Jordach celeron55, can i finish testing under 2.1.2 OGL?
12:40 celeron55 yes
12:41 Naked joined #minetest-dev
12:55 nore there is a problem with node pointing: if I point a node, and there is an entity *behind* the node, which would be pointed if the node was not there, it is the entity which is pointed and not the node
12:55 nore is that normal?
12:55 sfan5 no
12:56 nore then why does it happen? (btw, the entity is the same kind of entity as falling nodes, except that it doesn't fall)
12:57 nore i.e.: visual = "wield_item"
12:58 nore can you reproduce it?
12:59 celeron55 that has always been the case
13:00 celeron55 the pointing of things is calculated separately for nodes and objects, and objects take precedence
13:00 nore why?
13:00 celeron55 it just happens to be so
13:03 nore and couldn't be it be changed?
13:05 celeron55 probably can
13:05 celeron55 you need to combine two messy loops
13:06 nore nope: you get distance of nearest selected object
13:06 celeron55 oh whatever just fix it 8)
13:06 nore and if you find a node nearer than that distance, you have what you want
13:07 Jordach celeron55, you shouldn't merge it
13:07 Jordach it's decided to throw a shader compile error
13:11 Jordach http://i.imgur.com/RU3YJMD.png
13:11 Jordach with just the generate normalmaps enabled :P
13:17 ImQ009 joined #minetest-dev
13:18 celeron55 i guess it needs opengl 3 then
13:18 celeron55 well that doesn't matter as long as it works when the option is disabled
13:19 celeron55 but these should be documented and also possibly autodetected somehow
13:19 celeron55 RealBadAngel: what do you think
13:27 nore https://github.com/Novatux/minetest/commit/91923806a98fb5cc8e2bf6ed6a864b41337bb3e5 <-- any thoughts?
13:28 nore celeron55, ^
13:33 RealBadAngel celeron55,  i do have 1,8ghz amd 4600 series
13:33 RealBadAngel with all those options enabled im getting 40-50fps
13:33 RealBadAngel for 256x texture pack and generation enabled
13:34 celeron55 nore: test it with very large and very small objects near nodes; if it works for both, it's good
13:34 celeron55 RealBadAngel: i don't care about performance; i asked about compatibility problems
13:35 nore celeron55: the node is selected if the distance from camera to intersection with node is less than distance from camera to center of object
13:35 RealBadAngel celeron55, yes, there are problems, but in general not in shaders code
13:35 RealBadAngel but the drivers folks are using
13:36 nore and it is not possible to make better, since for objects, the one with the closest _center_ will be selected, not the closest intersection
13:36 RealBadAngel propertiary drivers are way faster
13:36 RealBadAngel thats why Thorvalds showed a finger to Nvidia
13:37 RealBadAngel also i can notice on the very same box that minetest runs way faster on windows
13:37 celeron55 nore: maybe that's good because otherwise you couldn't select small objects that are inside large ones... maybe
13:38 RealBadAngel another note: it runs WAY faster in full screen mode
13:38 celeron55 RealBadAngel: you are talking about completely random things
13:38 nore but is what I said ok about the selection objects/nodes, or do I need to change it?
13:39 celeron55 objects behind nodes are blocked while objects in front of nodes aren't; that was the idea, right?
13:39 celeron55 this means that objects inside nodes are blocked too; is that bad?
13:39 RealBadAngel celeron55, those are two issues on the very same code
13:39 celeron55 does some mod insert objects in a node and expects the player to be able to take them from there
13:40 RealBadAngel opengl is such a bitch that work different way depending on GPU
13:40 RealBadAngel and the drivers
13:40 nore yes
13:40 nore celeron55, it depends on the position of the object
13:41 BlockMen cauldrons are impossible then?!
13:41 RealBadAngel even if you restrict shaders code to a specific version, it does matter what drivers you are using
13:41 nore I don't think there are a lot of mods that do that...
13:41 RealBadAngel and then what brackets are allowed
13:41 nore BlockMen, what would you mean, cauldrons?
13:42 BlockMen nvm, i forgot that there is sth like nodeboxes^^
13:42 BlockMen ^nore
13:42 celeron55 what happens with nodeboxes that have a hole into them and there is an object inside?
13:43 nore celeron55: if the object is at the center of the node, and if you point the hole so that the intersection with the node is further than the center (that should be the case), then the object gets selected
13:43 celeron55 RealBadAngel: that doesn't mean we should simply disregard the user's problems
13:43 RealBadAngel no, ofc
13:44 RealBadAngel but i need feedback
13:44 RealBadAngel that means users tryin to run it
13:44 RealBadAngel and their error reports
13:44 celeron55 i would like it to be so that if shaders don't work, it asks the user whether he wants to disable shader features that don't work for him
13:45 celeron55 opengl's versioning is based on features, and the supported ones can be queried from the driver
13:45 celeron55 so minetest can disable what isn't supported
13:45 RealBadAngel atm shaders are compiled depending on chosen options
13:46 celeron55 nore: it's good then
13:47 RealBadAngel i am targeting glsl 1.20
13:47 celeron55 nore: i guess i'm responsible for this "pull request" so i can say that you can push it4
13:47 celeron55 -44
13:47 nore ok...
13:48 EvergreenTree joined #minetest-dev
13:48 EvergreenTree joined #minetest-dev
13:48 RealBadAngel celeron55, all the code named "shaders rework" was published way before i made a pull out of it
13:49 RealBadAngel and thx to feedback i have fixed many compability issues
13:49 RealBadAngel also old ones
13:51 eeew joined #minetest-dev
13:52 RealBadAngel and about the features, as i said, i decided to target 1.20
13:52 RealBadAngel that means even opensource drivers should work with it
13:53 RealBadAngel but opensource Nvidia refuses lighting for some unknown reason ;)
14:02 celeron55 RealBadAngel: should i push this?
14:05 VanessaE I need some help:  what would cause the server to create tons and tons and tons of players/xxxxxxxx.~mt  ?
14:05 VanessaE for each player on every world I run, all of a sudden, bam there they are
14:06 celeron55 it creates those for saving player files in a crash-resistant way
14:06 VanessaE doing so causes crashes. :P
14:06 VanessaE the server apparently can't tolerate the existence of those files.
14:07 celeron55 there shouldn't exist tons of them
14:07 VanessaE the whole dir was filled - every player had one on every single world
14:07 celeron55 only one for each crash where it has crashed in the middle of writing a player file
14:08 celeron55 then it's broken
14:08 celeron55 do you have a long path or some non-ascii characters in the path?
14:08 xyz sfan5: I told you, remember?
14:08 VanessaE http://pastebin.ubuntu.com/6678666/
14:08 VanessaE celeron55: nope, see above
14:09 VanessaE I came home from shopping to find that ^^^^^^^
14:09 xyz sfan5: you just need to root your device and then there's a workaround
14:09 xyz sfan5: and it's run-as that is broken afaik, not ndk-gdb
14:09 VanessaE the server was complaining of the old PlayerArgsEnd not found error, I investigated, found that ^^^^
14:10 VanessaE in fact all five were complaining, hence the use of find.
14:12 celeron55 how about if some logging of errors was added to filesys.cpp:707 so that we could have some idea why it fails in that case
14:12 VanessaE oh wait, I think I know what happened.
14:13 VanessaE looks like I may have ran out of space during the backup task
14:14 VanessaE still,
14:14 VanessaE looks like it revealed a corner case you might want to control for :)
14:14 celeron55 wtf
14:14 celeron55 oh
14:14 VanessaE "the backup task" == a script I run, not to the engine.
14:14 celeron55 so those .~mt files are empty?
14:15 VanessaE I didn't look.  deleted them and got the servers back up
14:15 celeron55 they probably were empty
14:16 celeron55 it tries to write one, and if it's empty, it just is like "oh that errored, i'll just do something else"; otherwise it moves it over the actual file (and thus never endds up with an actual file with half-written content)
14:16 celeron55 it should remove the file in the case of failing to write into it 8)
14:16 celeron55 (that it doesn't is a bug)
14:17 celeron55 wait what
14:17 celeron55 was the server trying to load a .~mt file?
14:17 celeron55 that's an another bug
14:18 sfan5 xyz: I found a workaround
14:18 VanessaE yes, it was
14:18 celeron55 i wonder if PilzAdam wants to fix these as he wrote the original 8)
14:18 VanessaE as soon as I deleted all *.~mt the servers were happy.
14:19 VanessaE the server loads *all* player files
14:19 VanessaE even if a player isn't signed on
14:19 celeron55 i know, i looked at the code
14:24 RealBadAngel celeron55, no, its not in the state for push. i was askin for testing that by devs
14:24 RealBadAngel the change only affects solids
14:25 RealBadAngel problem was nobody (of devs) wanted to try it
14:26 RealBadAngel if the generation code will work, i can make all the shaders use it
14:26 VanessaE celeron55: well that code needs fixed.  thanks to that, I can't even rely on timestamps on those files to be able to clear out old, dead accounts.
14:27 RealBadAngel but for testing is better to keep the changes in one
14:28 celeron55 RealBadAngel: you should make sure that the base-level shaders work with very low-end setups
14:28 celeron55 only if special features like that are taken into use, then it can require stuff like opengl 3
14:28 RealBadAngel i can achieve that only with feedback
14:29 celeron55 then ask Jordach always to test; he apparently has something non-fancy
14:29 RealBadAngel also lotsa other folks
14:29 RealBadAngel even VanessaE
14:30 RealBadAngel she had too new hardware, and problem with lighting popped up for her
14:30 VanessaE yup, because my driver doesn't clamp lighting values, it allows them to overexpose.
14:30 nore I have only OpenGL 2.1, so I can test it
14:30 RealBadAngel believe me, GLSL is a bitch to maintain
14:31 * sfan5 has some old hardware to test on
14:31 sfan5 but it will take 1 hour to compile
14:31 RealBadAngel lol
14:31 sfan5 literally 1 hour
14:31 VanessaE what is it, a RPi? :)
14:31 sfan5 no
14:32 nore for me it only takes 5 minutes, so I can test now if you need
14:32 RealBadAngel for me its enough to see logs to know whats wrong
14:32 sfan5 VanessaE: some old computer with an AMD processor (1 core) and a pretty old nvidia card
14:32 VanessaE RealBadAngel: and then...there's the glistening water shader ;)
14:32 VanessaE sfan5: ouch, must be damned old then
14:32 RealBadAngel the water will be reworked, you know
14:32 VanessaE RealBadAngel: yep I know
14:33 RealBadAngel water need to have surface, for christ sake
14:33 sfan5 does the rpi even support glsl?
14:33 VanessaE doubt it
14:33 nore RBA: reflections too?
14:33 RealBadAngel i do have reflections, refraction and surface code rdy
14:34 nore you can see the other nodes reflected in the water???
14:34 RealBadAngel ofc
14:34 * nore tests
14:34 RealBadAngel and the sunlight and moonlight
14:35 RealBadAngel i just need to split rendering into 2 passes
14:35 RealBadAngel up and below
14:35 RealBadAngel with result in texture shaders will do the rest
14:37 RealBadAngel http://www.youtube.com/watch?v=ax48OnXjDTE
14:37 RealBadAngel you can see water shaders here
14:38 celeron55 raspberry pi uses opengl es2
14:38 celeron55 it doesn't even display anythign without giving it shaders that emulate the fixed pipeline
14:38 celeron55 anything*
14:39 RealBadAngel quake3 works somehow on rpi
14:39 sapier joined #minetest-dev
14:41 RealBadAngel http://www.youtube.com/watch?v=YDPnqkZF6hQ
14:48 nore wow... I had to make a few tweaks, it killed my FPS, but the generate_normalmaps thing works!!! I couldn't see any reflections on the water however, and it stills use the normalmaps, so with bumpmapping, it is black...
14:48 PilzAdam joined #minetest-dev
14:52 xyz RealBadAngel: are water shaders in your repo already?
14:57 Exio4 sfan5, old nvidia card?
14:58 sfan5 yes
14:58 Exio4 nvidia igp?
14:58 Exio4 sempron?
14:58 Exio4 model name: AMD Sempron(tm) Processor LE-1200
14:58 Exio4 my actual cpu ;P
14:58 sfan5 not igp
14:58 sfan5 the mainboard has no vga output
14:58 Exio4 ah
14:58 sfan5 s/vga/video/
15:07 bas080 joined #minetest-dev
15:09 Megaf joined #minetest-dev
15:09 ImQ009 joined #minetest-dev
15:11 Yepoleb joined #minetest-dev
15:13 RealBadAngel xyz, no, water shaders are wip
15:15 RealBadAngel nore, the patch is about generate normal maps on the fly (or use overriding one)
15:15 nore yes, it works... (but not for liquids)
15:16 RealBadAngel yes, because only solids shader is modified to use it
15:16 RealBadAngel but you can try out one trick
15:16 RealBadAngel please create empy texture pack folder
15:16 RealBadAngel and copy there this image:
15:17 RealBadAngel http://i.imgur.com/epKsbgF.png
15:18 RealBadAngel rename it to: override_normal.png
15:18 RealBadAngel also i suggest turning off filtering
15:19 RealBadAngel this thingy will work for all the shaders actually
16:08 rubenwardy joined #minetest-dev
16:16 hmmmm joined #minetest-dev
16:19 bas080 joined #minetest-dev
16:38 Calinou joined #minetest-dev
16:38 Zeitgeist_ joined #minetest-dev
16:41 jin_xi joined #minetest-dev
16:55 rubenwardy joined #minetest-dev
17:12 Anchakor_ joined #minetest-dev
17:20 celeron55 xyz: are you interested in unifying this touch control stuff somehow sanely
17:21 xyz what do you mean?
17:22 celeron55 i'm not sure actually
17:22 celeron55 i
17:23 celeron55 i'm now using irrlicht's sdl2 backend quite succesfully in this
17:23 celeron55 or, well, one that i made...
17:25 celeron55 but basically i want to get as much of this upstream as possible
17:25 celeron55 upstream minetest (obviously)
17:25 xyz of what?
17:26 celeron55 code that allows one to actually play using touchscreen input
17:29 xyz well my code isn't android-dependent
17:29 xyz though at some places it is, but this can easily be changed
17:29 xyz you just need to send EET_MULTI_TOUCH_EVENT to it, and that's all
17:30 celeron55 you've mixed android-specific and other stuff in commits
17:30 celeron55 and the code is freeminer's so i can't copy files either
17:31 celeron55 i guess i'll try the commits
17:31 celeron55 i hate you for not committing to minetest though
17:32 xyz <3
17:33 celeron55 it's a serious pain and nothing progresses because freeminer isn't any better suited for continued development
17:33 xyz yeah sure
17:34 celeron55 well what
17:34 celeron55 do you suggest i close down minetest and start developing freeminer instead?
17:34 celeron55 yeah sure
17:35 xyz no, I didn't suggest that
18:03 Anchakor_ joined #minetest-dev
18:34 ImQ009 joined #minetest-dev
18:41 RealBadAngel celeron55, a few days ago i have bought new tablet, so i can test stuff
18:42 ImQ009 joined #minetest-dev
18:46 rubenwardy joined #minetest-dev
18:57 RealBadAngel2 joined #minetest-dev
19:00 sapier anyone another protocol to evaluate except udp tcp enet?
19:02 xyz don't think so
19:02 xyz how will we "evaluate" those?
19:04 sapier that's the open issue ;-) I have no real idea how to test, of course there are some numbers, e.g. cpu load, max download rate, rtt ... but none of them on its own is even close to a good indicator (at least if they're not extremely wrong)
19:05 daswort joined #minetest-dev
19:05 celeron55 well, there's only one real test: set up an identical server for both and play on them from everywhere in the world
19:06 celeron55 or, really, many servers with different connections...
19:06 sapier while this is best suggestion by now I don't really think this is a reliable test ;-)
19:07 sapier my current expectation is old udp support will loose any test (except of compatibility ;-) )
19:08 sapier tcp will win max bandwidth tests
19:08 sapier and enet hmm guess something in between
19:08 celeron55 enet could work best for crappy connections
19:09 celeron55 (overloaded crappy connections)
19:10 sapier you could be right about that
19:10 sapier oh I have to remove the useless connection lock prior starting tests
19:11 sapier enet might suffer off lack of send/receive separation ... but there could be other limitations for this too
19:12 sfan5 sapier: how about max. download speed when sending all media?
19:12 sapier guess it's hard to compete with tcp in this discipline
19:12 xyz honestly
19:13 xyz oh wait a moment
19:13 sapier and while that one is "first impression" it doesn't matter really much in game
19:14 xyz sapier: so is it fair enough to already compare your tcp_implementation_2 branch to minetest master?
19:15 sapier comparing to master isn't an option as master is utterly broken
19:16 xyz well I only did a small simple test
19:16 xyz just created a singleplayer game and placed some cobble blocks randomly
19:16 xyz and master won and tcp_implementation_2 sucked
19:16 xyz but if it's broken then I guess it's a bit unfair to compare, even if broken implementation wins
19:16 sapier can you be a little bit more precise?
19:17 xyz sapier: they disappear and reappear like on laggy servers
19:17 xyz I can record a video
19:19 sapier I believe you why would you lie about it ;-) question now is it is a silly bug in my implementation or a principle problem
19:19 sapier guess if you try to connect to a server with 200mb textures result will be different ;-)
19:20 sapier thats's why we need some more then a single test
19:24 sapier I'm not even sure flickering is indicator for low performance it might be result of asynchornous sending too
19:27 sapier old code handles sending and receiving in same thread so it's gonna process all currently received messages prior sending anything, this may result in queuing multiple map changes and processing them at once ... same way modifications to map are sent back to client in a single shot too
19:28 sapier you could verify this by using my fixed udp version, if it isn't limited by beeing to slow you should recognize same effect
19:28 xyz here's the video anyway http://a.pomf.se/cywoca.mkv
19:29 sapier how long do you have to wait for map to appear?
19:29 xyz the server_improvement one?
19:29 xyz just check the video
19:30 sapier ok I guess I need to download it instead of streaming
19:31 xyz can't reproduce this behavior in server_improvement branch
19:31 xyz works fine ther
19:32 sapier interesting
19:33 sapier is there a way to tell os to flush tcp buffers to network?
19:33 sapier thinking about a buffering issue
19:34 ShadowNinja TCP doesn't concentrate on latency, it concentrates on fast reliable delivery of bulk data.  Is it sent through TCP?
19:35 sapier that's why it's called tcp client ;)
19:38 sapier ok :-) I made the tcp client do the full buffering thing
19:40 sapier not even samba does it this way :-)
19:41 xyz so is it fixed now?
19:41 sapier not pushed yet
19:41 sapier I'm fixing the last bugs in enet right now give me a minute ... or two ;-)
19:43 xyz enet? you have this branch too?
19:44 sapier not yet pushed
19:52 kaeza joined #minetest-dev
19:56 kaeza it seems that `on_chat_message' calls all the registered callbacks regardless of what previous ones return. Is there a problem if such callbacks stop processing as soon as one of them returns some value?
19:57 sapier "some"? wasn't return value of various other callbacks bool? not quite sure about that either
19:58 Jordach joined #minetest-dev
19:59 kaeza as an example, I coded an "antiflood" mod, that disallows sending the same message twice in a row, by returning false from the callback, but the IRC mod is still getting passed the text, thus still flooding the channel
20:02 kaeza any insight on how to avoid this, without resorting to a heap of hackwork?
20:02 sapier flooding from minetest to irc? :-) quite ironic ;-)
20:02 sapier but anyhow how do you ensure your protection mod is first one to get the data?
20:03 sapier hmm "but anyhow" is german english ... just forget about that phrase
20:03 kaeza because antiflood (opt)depends on irc
20:03 kaeza I believe on_chat_message is FIFO, right?
20:03 sapier might be ... but there's no guarantee for it
20:17 ShadowBot joined #minetest-dev
20:19 kaeza oh derp
20:23 ShadowBot joined #minetest-dev
20:31 ShadowBot joined #minetest-dev
20:43 sapier1 joined #minetest-dev
20:44 Miner_48er joined #minetest-dev
20:47 Miner_48er joined #minetest-dev
20:48 john_minetest joined #minetest-dev
20:57 smoke_fumus joined #minetest-dev
21:16 ShadowBot joined #minetest-dev
21:19 eeew joined #minetest-dev
21:21 ShadowBot joined #minetest-dev
21:34 emptty joined #minetest-dev
22:04 eeew joined #minetest-dev
22:18 Naked joined #minetest-dev
22:21 Fury joined #minetest-dev
22:21 PilzAdam did someone fell asleep while releasing? half of http://dev.minetest.net/Releasing_Minetest isnt done
22:22 PilzAdam nobody pushed to stable-0.4 branch, mt_game isnt taged
22:22 PilzAdam the -dev suffix isnt enabled either
22:23 PilzAdam this is the second release that is only done half
22:28 naxthesurvivor_ joined #minetest-dev
22:41 diemartin joined #minetest-dev
22:50 diemartin joined #minetest-dev
22:57 iqualfragile_ joined #minetest-dev
22:58 celeron55 http://c55.me/random/2014-01/tscrot-2014-01-03_00-57-27.png
22:58 celeron55 so yeah, apparently ogles2 is broken in either minetest or irrlicht
22:59 celeron55 hard to tell
22:59 celeron55 it looks the same on the actual device
23:02 celeron55 it's like it drew only the first texture of a mesh or something
23:02 sfan5 i guess that is a bug in irrlicht
23:05 ShadowBot joined #minetest-dev
23:08 ShadowNinja And the serverlist is broken!!!
23:09 ShadowNinja The only usefull information that I got was that it got a error code of 0.
23:11 diemartin joined #minetest-dev
23:39 us_0gb joined #minetest-dev

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