Minetest logo

IRC log for #minetest-dev, 2015-01-11

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

All times shown according to UTC.

Time Nick Message
00:19 acerspyro joined #minetest-dev
00:24 Exio4 joined #minetest-dev
00:31 sapier left #minetest-dev
01:38 zat joined #minetest-dev
02:12 shadowzone joined #minetest-dev
02:32 hmmmm hey Zeno, aws wondering why you implemented gamma settings the way you did and not using IrrlichtDevice::setGammaRamp()
02:33 MinerDad joined #minetest-dev
02:46 OldCoder joined #minetest-dev
02:55 prozacgod joined #minetest-dev
03:13 gregorycu joined #minetest-dev
03:58 DFeniks joined #minetest-dev
05:16 Player_2 joined #minetest-dev
05:30 Miner_48er joined #minetest-dev
05:45 paramat joined #minetest-dev
05:48 paramat #2103 is perhaps ready for merge
05:48 ShadowBot https://github.com/minetest/minetest/issues/2103 -- Lua-api.txt: Document that paramtype=light is essential for a light source to spread its light by paramat
05:58 paramat hi hmmmm a simple plains biome (grasses, flowers) for mgv5/7 has just been added to MTgame, would you like me to remove builtin_biome?
05:59 hmmmm sure
06:01 hmmmm it would probably be a smarter idea to move builtin_biome to minimal
06:02 paramat okay
06:12 paramat currently mgv5 has mandatory caves carved in the base terrain loop, i think i should separate cave carving into it's own optional function and run that after biome surface material is added. currently biome surface material is placed in tunnels (oops)
06:15 hmmmm ahh, yeah, you should respect MG_CAVES...
06:22 apdapreturns joined #minetest-dev
06:23 MinerDad joined #minetest-dev
07:13 hmmmm heh hrmm
07:13 hmmmm anybody ever see a segfault caused by a bogus value in a vptr table?
07:14 hmmmm I tried compiling the irrlicht example #5 and that happens to me
07:15 gregorycu Are you sure if was a bogus value in the vptr table, or was it a bad "this" pointer?
07:15 hmmmm absolutely sure it was a bad vtable entry
07:16 hmmmm 3irr3gui13CGUIScrollBarD1Ev>    0x800d32d80 <_ZTv0_n48_N3irr3gui13CGUIScrollBarD
07:16 hmmmm 0Ev>
07:16 hmmmm 0x80102a3d0 <_ZTVN3irr3gui13CGUIScrollBarE+464>:        0xfffffffffffffe88
07:16 hmmmm 0xfffffffffffffe88
07:16 hmmmm erm
07:16 hmmmm that was meant to be more readable than that
07:16 gregorycu No idea
07:17 hmmmm crashes right here:  http://sourceforge.net/p/irrlicht/code/HEAD/tre​e/trunk/examples/05.UserInterface/main.cpp#l252
07:18 hmmmm well it's not a fluke since it happened with g++48 too
07:18 gregorycu No idea
07:19 gregorycu Umm....
07:19 gregorycu Is this a shared object, or a lib?
07:20 hmmmm hah
07:20 hmmmm well let's put it this way, it's not a compiler bug
07:21 gregorycu You'd get this type of shit, if you were using a shared lib (dynamic loaded)
07:21 gregorycu Where the shared lib didn't match the headers you used
07:22 hmmmm ahh
07:22 hmmmm let's try removing the "-I/../../include"
07:23 hmmmm and it works
07:23 hmmmm you're awesome :)  good job recognizing that
07:23 gregorycu It's my job
07:24 gregorycu Which I have to go back to tomorrow after 20 days off
07:24 hmmmm :(
07:24 hmmmm is it soul crushing?
07:25 gregorycu What will be soul crushing is all the emails, and also trying to get back on track with what I was doing 20 days ago
07:25 MinerDad joined #minetest-dev
07:25 gregorycu My memory is shit
07:26 hmmmm going to work 'tomorrow', eh?  you sound australian
07:26 hmmmm aha, hostname confirms
07:26 gregorycu I certainly am
07:27 gregorycu Zeno and I make sure the Australian quota is met in this project
07:27 hmmmm you australians are always a day ahead of us americans
07:27 hmmmm literally :p
07:28 gregorycu If it makes you feel better, when the Armageddon comes, we'll be the first to go
07:37 Hunterz joined #minetest-dev
07:44 paramat okay hmmmm #2036 is tested and ready for review
07:44 ShadowBot https://github.com/minetest/minetest/issues/2036 -- Builtin: Remove builtin_biome.lua. Minimal: Add simple biome to default/mapgen.lua. Remove indev mapgen ores by paramat
07:52 Calinou joined #minetest-dev
07:53 paramat also #2103 if my documentation of paramtype="light" is correct
07:53 ShadowBot https://github.com/minetest/minetest/issues/2103 -- Lua-api.txt: Document that paramtype=light is essential for a light source to spread its light by paramat
07:56 nore joined #minetest-dev
08:11 paramat nore mgv5/v7 and the biome API are not stable yet, perhaps they will be at next stable release. however i'm about to start adding optional snow biomes to mgv6 if you and sfan5 still want them
08:13 nore I do :)
08:14 nore it would be good to have pines and snow by default in _game...
08:15 nore btw, if you make snow biomes, I suggest you to make two different ones: one with few trees and one like a forest
08:18 paramat yes i would like to have taiga and tundra: cold and wet, cold and dry
08:19 paramat as a mirror image of jungle and desert
08:21 paramat also, i am working on a full set of biomes for v5/v7 with new tree schematics, hopefully for next stable. what is in MTgame currently is just temporary
08:22 nore oh, and how have you done with the trees? did you generate dozens of pinetrees and then saved them as as many schematics?
08:29 paramat haven't started tree schematics yet. most will use the randomised node feature of schematics, but snowy pines may done differently
08:29 paramat .. because the per-node randomisation cannot place a snow node on every leaf node
08:30 Calinou +nore> btw, if you make snow biomes, I suggest you to make two different ones: one with few trees and one like a forest
08:30 Calinou this is very important
08:31 paramat taiga can be split into forest and clearings by the tree perlin noise, i could even use that noise to vary tree density?
08:31 selat joined #minetest-dev
08:32 Calinou yes, noise to vary tree density would be good
08:32 Calinou although make it so that normal forests behave this way too
08:32 paramat yes vary tree density by noise, i find that essential to look good
08:32 Calinou (for consistency)
08:32 paramat that might break old worlds a little though (not much)
08:33 paramat actually the discontinuity won't be ugly..
08:33 paramat so yes i agree
08:34 Calinou so that we have highly forested areas, scarcely forested ones, and not-at-all forested ones
08:34 Calinou forests wouldn't be a separate biome anymore, which is good for simplicity
08:35 Calinou we'd have just plains/desert/tundra
08:35 Calinou and trees/cacti on them
08:38 paramat tundra is apparently cold, dry and barren, no trees, perhaps a new 'permafrost' node. then also i plan to add icesheet only in the very coldest areas
08:39 paramat or, tundra could be rocky, no dirt at all, just thin snow
08:42 Calinou keep dirt
08:43 Calinou we could have ice lakes sometimes
08:46 paramat yes icesheet on water in extreme cold
08:59 kilbith joined #minetest-dev
09:04 paramat left #minetest-dev
09:05 jluc joined #minetest-dev
09:14 MinerDad joined #minetest-dev
09:14 Krock joined #minetest-dev
09:20 Krock why is that function there? https://github.com/minetest/minetes​t/blob/master/src/guiTable.cpp#L326
09:37 ImQ009 joined #minetest-dev
09:47 kilbith joined #minetest-dev
10:08 sapier joined #minetest-dev
10:16 paramat joined #minetest-dev
10:19 paramat nore, sfan5 https://github.com/minetest/minetest_game/pull/399
10:19 paramat left #minetest-dev
10:26 cib0 joined #minetest-dev
10:28 nore game#399 was the fastest-merged PR :)
10:28 ShadowBot https://github.com/minetes​t/minetest_game/issues/399 -- Return to original water sink speed for player by paramat
10:28 sfan5 yeah
10:33 Calinou the eleventh world war is over!!
10:35 sapier main.cpp L1612 ... things like that should be hidden within the porting layer ... or the function name is wrong if this can be done anywhere else then on x too
10:40 puhfa hey, you guys aware that memory is leaking something fierce especially when handling bunch of entities?
10:40 puhfa managed to actually crash my system when both 4 gb ram and pagefile where filled with cobble after only 30 mins of running the client
10:41 puhfa 0.4.11-64 windows build from the website
10:46 ImQ009 joined #minetest-dev
10:56 sfan5 ShadowNinja: having blocker and high priority labels on an issue doesn't make sense, blockers are always high priority
10:59 sapier sfan5 can you explain to me why fixing a feature not even beeing officially supported is not "low priority" ... and moreover why (you) removed the "low priority" flag in total but kept medium and high?
11:00 sfan5 sapier: every issue that is not high or medium priority is automatically low priority
11:00 sfan5 we don't need a label for that
11:00 sapier because you believe we don't need it?
11:00 sfan5 what do we need it for?
11:01 sapier if you don't have it consequence is everything not beeing medium or high will be LOW
11:01 sfan5 what should the default priority for issues be?
11:02 nore medium priority?
11:02 sapier imho low is a flag for things that are a bug but are at "glitch" level like "would be nice to fix but doesn't actually do any harm"
11:02 sapier e.g. a invalid warning message for example
11:03 sfan5 maybe we should remove the medium label and add the low label back
11:03 sapier maybe at least I don't really have feeling what medium is supposed to be
11:04 sapier we can't add a desctiption to labels on github can we?
11:04 sfan5 no
11:05 sapier guess it'd make things way more easy if we knew what a label was supposed to mark
11:05 sapier I'd suggest keeping the medium label as it's at some issues right now but not using it any longer
11:30 exio4 joined #minetest-dev
11:49 gregorycu The default priority is null
11:49 gregorycu Which means it hasn't been assigned a priority
11:49 SudoAptGetPlay joined #minetest-dev
11:49 gregorycu Surely you want to explicitly categorise things, and have some way of knowing which things have not been categorised
11:50 SudoAptGetPlay Hi, how can I register someone pointing at an other player (to fire an hud element)
12:11 MinetestForFun joined #minetest-dev
12:12 puhfa 3 Gb in less than 20 min :|
12:12 puhfa so restart is needed twice in an hour
12:12 puhfa is this a known issue?
12:12 sapier puhfa: what version are you using?
12:13 puhfa now tried the previous one, 0.4.10 stable
12:13 puhfa cant really say if the latest stable is any better
12:13 sapier so 0.4.10 uses that much memory?
12:13 puhfa yes. but the thing is, i am deliberately spamming this entities
12:14 puhfa using a pipeworks machine
12:14 puhfa with entities*
12:14 sapier well in this case stop spamming ;-)
12:14 kahrl sounds like a case of "doctor, it hurts when I do this" :P
12:14 puhfa the leak will still be there, stalking. what if it attacks me when im asleep? :
12:15 sapier kahrl: true yet we could have issues with handling a entitiy overload scenario too
12:15 puhfa and the machine is no lets-break-all-clients thing, just an ore processor
12:16 sapier pipeworks merely abuses some of minetests mechanisms it's not a surprise it trigger issues never happening with regular usage
12:16 puhfa id say using pipeworks counts as regular usage these days
12:16 sapier I don't
12:17 puhfa anyway, i am getting serialization errors once in a while here, might be related
12:17 gregorycu Hmm...
12:17 gregorycu What does pipeworks spam?
12:17 sapier pipeworks main issue is it doesn't handle minetests activity range
12:18 sapier activity range concept is by design incompatible to what pipeworks does
12:18 gregorycu Maybe so
12:18 gregorycu But what does it spam?
12:18 puhfa oh, forgot to mention, i keep the map block always loaded when i use the tubes
12:18 SudoAptGetPlay left #minetest-dev
12:18 puhfa might help narrowing it down
12:18 sapier which is even worse because that feature is flaged "use at your own risk"
12:18 puhfa gregorycu: i am just using it a lot
12:20 gregorycu So, there is no reason for the memory usage to increase
12:20 puhfa so does this mean youre not interested in trying to fix it, even though its a memory leak?
12:20 sapier well if you force load a block and add more and more entities memory increases that's quite natural
12:20 gregorycu Is that what is happening here?
12:21 puhfa um, the old objects should disappear but they still stay in memory
12:21 nore pipeworks tries not to spam unloaded blocks with entities
12:21 puhfa they arent freed properly
12:21 gregorycu The definition of a memory leak does not included entities that deliberately hang around
12:21 puhfa no they dont. they spawn, hang in there for a while and then supposedly disappear. except they wont
12:21 sapier maybe, I guess at some point you're startung to build up client <-> server message queues too as the number of entities causes a lot of communication
12:22 gregorycu Right, but let's not quickly assume it's a dodgy mod
12:22 puhfa basically its just transporting items from one inventory to another via a short tube
12:22 sapier this is a guess, but I'd not be surprised if you could cause a runnaway effect
12:23 sapier btw if you overload entity system client server could get out of sync too
12:23 puhfa that could be the cause, yes. someone probably should test running pipeworks machines with little traffic but for a long time
12:23 sapier maybe client thinks a entity dropped on server is still present resulting in it hanging around on client inifinite
12:24 puhfa tried some noclipping, no visible entities
12:24 sapier puhfa: yes maybe, you just have to find someone willing to fix pipeworks bugs
12:24 kahrl so there are a lot of places that could be the cause
12:24 kahrl I think this means this needs to be investigated with massif (which I have to admit to have no clue how to use)
12:25 gregorycu wtf is that?
12:25 kahrl gregorycu: a part of valgrind
12:25 sapier massif is a heap profiler
12:25 gregorycu ahh ok
12:26 sapier you can find where "not lost" memory is stored
12:26 gregorycu I'll grab pipeworks, and have a play
12:29 puhfa sapier: i am not saying pipeworks is perfect, i just think i found a memory leak while using it
12:29 puhfa no matter how twisted the mod is, it causes the engine to misbehave
12:30 sapier android sdk is crap ... if you wanna use version specific styles you have to increase target api to highest style you wanna support ... resulting in not having any check about your specified min api level still would work
12:30 puhfa anyway, in summary: lots of entities spawned and destroyed, no chunks loaded or unloaded, lots of inventory movements
12:30 sapier puhfa: a memory leak is lost memory ;-) I don't think we have lost memory here
12:31 gregorycu What makes you think that
12:31 puhfa uh, okay
12:31 puhfa anyways, allocated memory is not freed
12:31 gregorycu sapier doesn't speak for everyone
12:31 puhfa this is what is happening
12:31 puhfa whatever you want to call it
12:31 sapier I did a lot of memory leak checking the last week I found and fixed  a few, none related to entities at all
12:31 kahrl well, let's see what massif is showing
12:32 sapier I'm with kahrl I guess the memory is just stalled somewhere
12:32 gregorycu Stalled?
12:32 sapier yet It's still a hypothesis nothing proven
12:32 gregorycu What is stalled memory?
12:32 puhfa wouldnt that be a memory leak then? :)
12:32 sapier nope
12:33 kahrl gregorycu: meaning there's still a pointer to it (e.g. it's in a queue), but the server can't keep up with processing that queue
12:33 sapier by definition a leak is unreachable memory
12:33 gregorycu It doesn't have to be unreachable
12:33 gregorycu It just has to be never used again
12:33 sapier in order to be a leak it has to be unreachable
12:33 puhfa well if the client doenst ever use that pointer, seems pretty unreachable to me
12:33 sapier effect of stalled and unreachable is same of course yet it's a big difference
12:33 puhfa okay
12:34 gregorycu That's how some algos find memory leaks
12:34 sapier it's easy to find unreachable memory but stalled memory may require certain circumstances to happen
12:34 sapier e.g. massive entity overload by pipeworks ;-)
12:35 gregorycu The engine should be robust to handle this usecase
12:35 sapier well maybe the engine is if you don't enable hacks like forceloading blocks
12:35 kahrl an easy test would be: let the pipeworks machine do its thing for a while, then stop it, and see if the memory usage goes down
12:35 gregorycu If puhfa isn't going out of his way, and bad shit is happening...
12:36 gregorycu Is there a pipeworks tutorial somewhere?
12:36 sapier yes kahrl and disable forceloading if the machine is that small it's not required
12:37 gregorycu What does forceloading do?
12:37 gregorycu I thought it simulated a player standing somewhere
12:37 sapier disable activity mechanism for a certain block keeping it active
12:37 gregorycu Why should that have an impact on anything?
12:37 sapier gregorycu:  in theory yes but I don't know how it's been implemented
12:38 gregorycu Indeed, ok, give me a few min
12:38 sapier if there's  a bug in this hack It could interfer with regular entity handling
12:38 T4im1 joined #minetest-dev
12:39 Krock interesting. FPS drop frm 40 to 20 when placing a node but drawtype stays at 3
12:39 Krock s/drawtype/drawtime
12:40 T4im joined #minetest-dev
12:40 sapier I don't believe fps drop if you place a single node
12:40 Krock it actually happens
12:41 sapier what os?
12:41 Krock windows 5.1
12:41 Krock screenshots coming
12:41 sapier btw drawtime does only contain the time to draw the scene all logics is done out of this timer
12:41 crazyR joined #minetest-dev
12:42 sapier fps == 1s / ((time to do logics) + drawtime)
12:42 sapier -s
12:42 jluc joined #minetest-dev
12:43 sapier yet placing a single node should cause any noticeable change
12:43 Krock http://imgur.com/TwRu0Vs,F05GWPr,GmRepng
12:44 Krock 1) normal 2) placed 1 node 3) places 2 nodes
12:46 Krock interesting. it's gone now when the older chunks got unloaded by the client
12:46 sapier did you wayt for fps to settle down prior taking the first screenshot?
12:46 sapier grrr what the heck am I typing
12:47 puhfa hmm, thats happened to me too
12:47 puhfa mainly if i am placing the nodes very high
12:47 puhfa i suspect it is just generating shadows
12:47 sapier Krock what machine are you running this at?
12:47 Krock Athlon 64 4000+ 2.4 GHz single core, 2 GB ram
12:47 puhfa Krock: is the ground block below you loaded? because then ive seen this too
12:47 sapier drawtime doesn't change so if there ain't a bug in drawtime calculation fps drop is caused by some server action
12:48 Krock Radeon X300 graphics card
12:48 sapier hmm windows 5.1 is windows xp am I right?
12:48 Krock puhfa, now they're unloaded and problem is solved
12:48 Krock sapier, yes
12:49 sapier you're lucky on radeon x200 minetest won't even start ;-)
12:49 Krock how come?
12:49 sapier I don't know
12:49 puhfa it happens to me if the blocks below me are loaded and theres plenty of distance (height)
12:49 puhfa so shadows?
12:49 sapier tried it once on my old laptop ... never took time to find out
12:50 sapier krock's drop isn't related to graphics
12:50 fz72 joined #minetest-dev
12:50 puhfa no, i mean lighting calculations
12:50 Krock it looks like something is calculated.. and that something uses much CPU
12:50 Krock yeah. maybe lighting
12:50 sapier looks like unloading blocks takes some cpu time, enough to cause a fps drop ... light calculations might be a reason yes
12:51 MinerDad joined #minetest-dev
12:52 Kodexky joined #minetest-dev
12:52 Krock so if lightening would be calculated with the GPU, I would get drawtype spikes
12:53 sapier in theory yes, but gpu can't calculate the node light
12:53 sapier well maybe using gpgpu capabilities
12:54 sapier but not as in 3d graphics lighting
12:54 Kodexky Hi sapier.
12:54 Kodexky Nice to meet you.
12:54 sapier thanks Kodexky
12:55 kahrl well I definitely see some optimization opportunities in Map::propagateSunlight
12:55 sapier do you wanna do gpgpu coding kahrl? ;-)
12:55 Kodexky please forgive, but on 25235a4 shouldn't BS be 1 for that to work?
12:55 kahrl huh why
12:56 kahrl I mean the getBlockNoCreate call is only needed at the start and when y is crossing a mapblock boundary
12:56 kahrl and should be changed to getBlockNoCreateNoEx as well
12:56 sapier oops
12:56 sapier you're right Kodexky
12:57 Kodexky ok, i thought i was more too sleepy
12:57 Kodexky -more
12:58 sapier fixed
12:58 Kodexky cool :) now to fix the sleepy part...
12:59 celeron55 Kodexky: umm hello, i guess you came here due to my message
13:00 Kodexky hi celeron55, yes in part.
13:03 Kodexky sorry just remembered the forum and went checking, im very slow right now.
13:04 celeron55 it looks like we have many poeple who are working on some android specific parts of minetest, i wonder of people feel okay with the current amount of coordination or whether there should be clearer goals
13:04 Kodexky being honest i'm not too comfortable on irc, im just trying to like it
13:05 celeron55 this channel is often a complete chaos, i totally get it if someone doesn't like it
13:06 celeron55 Kodexky: what do you prefer the most over IRC? there is a bit of ongoing discussion over this too
13:08 sapier who else is working on android celeron55?
13:08 Kodexky im quite fine using mail, people breaths and thinks a little more before submiting.
13:08 Kodexky the delay is the problem of course.
13:09 sapier I'd be more concerned about the amount of mails Kodexky
13:09 celeron55 sapier: Kodexky has done some pull requests, and one other guy on the forum that i never heard of before seems to have a lot of work put to it
13:09 sapier guess I should have a look at forum?
13:10 celeron55 this reply: https://forum.minetest.net/vi​ewtopic.php?p=167371#p167371
13:11 celeron55 how do we combine these efforts to give the best result to users
13:11 sfan5 that guy released minetest on ouya
13:12 celeron55 well Kodexky's pull requests are perfectly fine - what i'm concerned in that case is for example duplicate work
13:13 sapier well I spent about 6 hours of work for trash yesterday because of ppl not beeing able to actually test prior complaining do you really assume duplicated work is an issue?
13:13 celeron55 well true, it generally isn't
13:14 sapier where's that pull request?
13:14 sfan5 sapier: if you're talking about the font stuff, it looks like it wasn't really clear what the problem was and what needs to be changed
13:14 Kodexky lets make a list of current and choose from there
13:14 celeron55 i'm not referring to any existing problem; also all those PRs have been merged
13:14 sapier the menu mockup is interesting but I'm not sure if it'd work on a smartphone with this orientation
13:15 celeron55 do we just tell to ezraanderson that his patches are welcome and he should submit PRs of them?
13:15 celeron55 or should we make a list of goals for the android port
13:15 celeron55 and then say PRs for those will be accepted
13:15 sapier some goals would at least help to make us move towards same direction
13:16 sapier we've got a pr for most of them already
13:17 Kodexky both, id really like to take a look at what ezraanderson has done, and a roadmap stating requirements as a priority.
13:17 celeron55 a wiki page to describe goals and constraints for android development maybe
13:18 celeron55 (constraints like what techniques are acceptable for implementing better menus)
13:18 sapier Kodexky: if you're talking about that post the only major things not yet merged is the hw image scaling and the help system
13:18 gregorycu puhfa: You there?
13:19 Kodexky just curiosity. :)
13:19 sapier for hw scaling I suggest doing this  little bit more precise just render at 320x480 for everyone seems crude to me :-) there are plenty of devices capable of running mt on native resolution
13:20 Kodexky dont make it mandatory.
13:20 puhfa gregorycu: yeah
13:20 puhfa sup?
13:20 Kodexky and low requirements option.
13:20 gregorycu puhfa: How do I make this continually output items?
13:21 Kodexky many games have that
13:21 gregorycu I have two chests linked by two pipes
13:21 puhfa um, you could use blinky plants from mesecons
13:21 sapier Kodexky: that's what I meant with "more precise"
13:21 gregorycu with two filter-injector
13:21 gregorycu blinky plants... ok
13:21 Kodexky ok
13:22 puhfa i was using technic machines that auto-inject their contents
13:22 sapier yet his patch doesn't contain any support do en/disable this so we'd have to add that interface prior merge
13:22 Amaz joined #minetest-dev
13:22 gregorycu I was searching for "clock" and "pulsar", I should have searched for "blinky plant"
13:22 sapier guess that's a minor issue
13:22 celeron55 sapier, Kodexky: would you contribute to an android roadmap wiki page if there was one?
13:22 puhfa gregorycu: logical, aint it? :)
13:22 sapier celeron55 every now and then yes ... but I don't like wiki very well I always forget about my password there
13:23 celeron55 can you think of something better
13:23 celeron55 IRC certainly isn't good for keeping track of goals and progress
13:23 celeron55 just milestoned issues on github?
13:23 sapier nope ... it's not a wiki problem but a "i need 5 different passwords for minetest alone" issue ;-)
13:24 sapier actuall it's 6
13:24 sapier wiki forum github irc android keystore android cert
13:24 puhfa gregorycu: btw, i had like 32 machines injecting stuff every abm, so thats the scale we are talking about. might not happen with a single chest+blinky config
13:24 sapier hmm what's number 6 ...
13:25 gregorycu sweet jesus...
13:25 sapier lol
13:25 puhfa what. i told you guys i was spamming the cobble out of it
13:25 gregorycu Why is it a fucking plant
13:25 sapier causing overload intentionally and complain about memory usage later ;-)
13:25 puhfa 32 actually makes sense. what with the 32 inventory slots and the like
13:25 sapier puhfa: sounds more like a benchmark then regular usage
13:26 Kodexky im for the milestones on github
13:26 puhfa i like to push my limits...
13:26 gregorycu 32 is nothing
13:26 puhfa anyway, still, even when theres an overload, it should free the entities
13:26 sapier well guy guess you found out you're over it
13:26 puhfa or perhaps have some kind of timeout that auto-destroys them
13:26 celeron55 sapier: are you fine with milestone issues on github?
13:26 sapier yes would be a good solution too
13:27 sapier I have to do some real life work now I'll be back in about 2-3h
13:27 sapier left #minetest-dev
13:27 celeron55 okay, that can be the way to write down goals for android then
13:29 younishd joined #minetest-dev
13:29 celeron55 anyway, there doesn't seem to be any real issues so i'm not going to mess with this more
13:30 celeron55 if everyone is too lazy write down any goals, there are plenty of small issues to be solved in any case
13:32 Kodexky let me think it again, after talking to my pillow... i just burned myself with a cigarette :s
13:32 Kodexky not quite functional right now.
13:33 celeron55 s/too lazy write down/too lazy to write down/
13:35 Kodexky well, it's been a pleasure.  guess we can go with the details later.
13:36 Kodexky now i need to rest.
13:36 Kodexky see ya
13:37 celeron55 good night, or something 8)
13:47 gregorycu puhfa: My initial thoughts are that there is a leak here
13:48 gregorycu The only other think worth mentioning is that I have a CPU kinda pegged at 100%
13:48 gregorycu Or near abouts
13:50 puhfa high cpu usage kinda makes sense, lots of entities to handle and i am not worried about that
13:51 puhfa anyway, when the entities disappear from view, they should leave the ram as well
13:51 gregorycu Well, if the CPU usage means that there is something queueing somewhere...
13:51 gregorycu The entities are particles if I am not mistaken
13:51 puhfa did you get any serialization errors due to having too many entities around?
13:52 gregorycu No, but I only have about 10 pumps going
13:52 puhfa i think i actually had ~64 nodes injecting stuff constantly
13:52 puhfa but those entities were supposed to be short-lived
13:53 puhfa i have seen this happen before and so have others
13:53 puhfa i brought it up because this was the first time my system actually crashed because of it
13:53 T4im it always seemed to me, that the cpu only spikes around the time it tries to swap around gigabytes of that entityleaking… i.e. more as a side effect…
13:53 T4im but I might be wrong
13:54 T4im it is a very common occurrence, yes
13:54 puhfa well, when using certain mods at least
13:54 T4im and not just on technic runnign servers, its just most noticeable there
13:55 T4im actually pipeworks running server we should say
13:58 FR^2 joined #minetest-dev
13:58 gregorycu hmmm...
13:58 gregorycu Could be particles
13:59 gregorycu I'll do tests tomorrow
13:59 puhfa i dunno. lua-side they can have all sorts of properties stamped onto them
13:59 puhfa oh
13:59 puhfa nevermind
13:59 kilbith sfan5: what's that font name ? http://i.imgur.com/7EcOO.png
13:59 sfan5 no idea
14:00 sfan5 jordach made that screenshot
14:00 nore joined #minetest-dev
14:06 roniz joined #minetest-dev
14:07 fz72 I found a memory bug: when I leave a world the memory doesn't get free
14:55 fz72 joined #minetest-dev
15:04 casimir joined #minetest-dev
15:05 casimir It seems with the new sorting the server list is easy to manipulate.
15:07 sfan5 casimir: the old sorting was as easy to manipulate too
15:07 proller joined #minetest-dev
15:08 casimir Yes. But now I made it up to the third place on the list, without effort.
15:10 sfan5 what do you mean
15:12 selat joined #minetest-dev
15:13 casimir Before my server was very unknown, with only one or two players at max. Now it almost hits max clients for no other reason than the sorting.
15:14 casimir 1. It seems my server is in the same data center (or near by) as the masterserver. ping = 1
15:15 casimir 2. Second guests are disallowed, so only full players get counted, and they give more points.
15:15 casimir 3. I increased the server time just to test if it is possible to manipulate the list. ;)
15:17 casimir Then choose your max clients to be 127 and let it run for at least an hour.
15:18 sfan5 regarding 1) servers.minetest.net is in amsterdam at waveride/edis, your server is in amsterdam at digitalocean
15:18 sfan5 2 is expected behaviour
15:18 casimir 3 Should be changed.
15:19 sfan5 why?
15:19 casimir It just profits those who know about it. In the long run every one will increase the game_time to be above 8 months.
15:21 Wayward_One joined #minetest-dev
15:21 sfan5 the plan was to allow older servers (more stable) to be at the top
15:22 sfan5 the plan wasn't that server owners increase the age of their world by editing some files
15:23 casimir Yes. That's the problem.
15:23 casimir (btw. I will turn mine back to the normal time)
15:23 sfan5 how do you suggest to solve this
15:24 casimir Just don't count the time.
15:27 sfan5 casimir: 64 bytes from 178.62.146.72: icmp_seq=1 ttl=60 time=0.954 ms the ping isn't even 1ms
15:28 sfan5 casimir: take a look at http://servers.minetest.net it doesn't count the time anymore
15:28 Wayward_One would it be possible to measure how long a server has been around externally? like say, how long it's been on the server list?
15:30 Wayward_One externally meaning not affected by someone editing the age of their world
15:30 sfan5 yes
15:30 sfan5 but that would require extra code
15:31 Wayward_One hmm
15:35 kaeza joined #minetest-dev
15:40 sapier joined #minetest-dev
15:49 VanessaE hmmmm:  two more reports of corrupted blocks for you,
15:49 VanessaE [01-11 10:38] <Gael-de-Sailly> http://i.imgur.com/vJsYIRx.png
15:49 VanessaE [01-11 10:49] <lordawe> I DO see cubes: http://i.cubeupload.com/tmTtli.png
15:50 casimir Thank you sfan5, it now went down to 9th place. Amazing to see how a relative small change leads to ab bigger change in a chaotic system. (The chaos part here are the players.)
15:50 sfan5 well
15:50 sfan5 it depends when you looked at the serverlist
15:51 sfan5 i tweaked it a bit
15:51 casimir what else have you done?
15:51 sfan5 casimir: http://sprunge.us/UfXg?diff
15:53 sfan5 hm
15:56 sfan5 made another change
15:57 sfan5 shouldn't be so bad for servers with lower average_players now
15:58 sfan5 hm
15:58 sfan5 still not ideal
15:59 sapier sfan5 "It's broken" ;-P
15:59 sfan5 ?
16:01 crazyR joined #minetest-dev
16:01 casimir Keep it simple
16:03 sfan5 yup
16:03 sfan5 i think this is good
16:07 kahrl so those windmill clusters
16:07 kahrl are they caused by 8621e6de5d?
16:09 cib0 joined #minetest-dev
16:09 kahrl I don't see where new_data is initialized in addArea()
16:10 kahrl the loops just copy the old data but where is the rest of new_data initialized
16:10 sapier isn't the whole area initialize by memset? at least do I rememer it beeing that way
16:11 kahrl there is no memset on new_data
16:11 sapier strange
16:11 kahrl there is one for new_flags but that doesn't matter
16:12 kahrl perhaps PilzAdam was right ;) http://irc.minetest.ru/minet​est-dev/2014-12-24#i_4077862
16:13 sapier well the area copy/add code is tricky I learned this the hard way too
16:14 sapier what's the voxel.cpp change??????
16:14 sapier who did agree to this?
16:15 sapier if compiler does align to 4/8 for arrays this might fail completely
16:15 kahrl I don't remember who did
16:16 sapier I've to check this once again it's so crude I'm not sure it may work by accident
16:16 sapier at least on some compilers
16:19 sapier no what that's just wrong
16:19 sapier a MapNode constuctor ain't called if no MapNode element is created
16:21 kahrl well the idea was to avoid calling the MapNode constructor
16:21 kahrl because calling it is slower than not calling it
16:21 kahrl but not calling it is undefined behaviour, so...
16:21 sapier well if this is supposed to create a list of pointers to MapNodes constuctor won't be called anyway
16:21 sapier if it's supposed to create an array of MapNodes the new code doesn't allocate enough memory
16:22 sapier because sizeof(pointer) != sizeof(MapNode)
16:22 kahrl sizeof(*new_data) == sizeof(MapNode)
16:23 sapier wait dereferenciate a pointer for sizeof? does this work?
16:23 kahrl yes
16:23 sapier stange way of writing code
16:24 kahrl it's less prone to being missed in case someone changes the type that is pointed to
16:24 sapier well if this conforms standard it'd be correct
16:24 sapier maybe but way more hard to read
16:24 kahrl I don't find it hard to read
16:25 sapier what about a memset(new_data,0, new_size * sizeof(*new_data))
16:25 kahrl it's an idiom in C (where you actually need to pass something like that to malloc often)
16:26 MinerDad joined #minetest-dev
16:26 kahrl shouldn't it be initialized to CONTENT_IGNORE?
16:26 sapier hope 0 is CONTENT_IGNORE
16:26 kahrl though it was initialized to CONTENT_AIR before, I don't know if and how that worked
16:26 kahrl nope, 0 is whatever is registered first by a mod
16:27 kahrl CONTENT_IGNORE is 127
16:27 sapier ok then this optimization is just crap
16:27 sapier we need to have defined values in there. If we can't use a memset we can use the constructor too
16:27 kahrl agreed
16:29 sapier As I said it's not first time I made a similar mistake too
16:33 kahrl anyway let's wait for Zeno`, maybe he can explain how it works :)
16:33 sapier yes it's broken and wont get different if we don't do anything maybe we miss something
16:35 VanessaE fwiw those big clusters of errant nodes aren't limited to windmills
16:35 VanessaE if that helps any
16:36 VanessaE here's the one that happened the other day on one of my servers:  http://digitalaudioconcepts.com/vanessa/h​obbies/minetest/screenshots/Screenshot%20​-%2001102015%20-%2001%3a03%3a25%20AM.png
16:36 VanessaE you can see where normal mapgen ops even took place after the errant mapblocks were created (the trees and plants on top, and even mudflow)
16:43 hmmmm joined #minetest-dev
16:50 Calinou joined #minetest-dev
17:16 sapier celeron55: I can't implement the upscaling feature, I'd have to touch formspec code and I'm not gonna do this ahain
17:22 celeron55 we seriously need to rework or replace formspecs to be based on a proper design with clearly defined behavior
17:22 celeron55 this is unbearable
17:24 celeron55 https://forum.minetest.net/vi​ewtopic.php?p=167434#p167434
17:24 celeron55 i could try to hook these guys up to this task, assuming they're competent enough
17:24 celeron55 they want a proper UI themselves too in any case
17:25 celeron55 it's just that whether they are willing to spend 75% of their time thinking about networking when they could just hack something that works specifically for their project
17:26 sapier I'm not gonna do this celeron55 I've already had enough trouble, I'm gonna support in designing it but I'll not code it
17:26 celeron55 i think i'll have to ask what these guys have in mind
17:28 sapier for those hw scaling the issue is formspec uses the actual screen size to be placed which is faked by the scaling change
17:30 kilbith http://kenney.itch.io/voxus
17:31 kilbith (no mention of Minetest yet...)
17:31 sapier I'm not sure if those guys would be right they're designing their game I'd be worried about them having a limited point of view
17:36 sapier btw I don't even think formspec language is that bad, it's some historic inconsistencys but for a gui system not supporting dynamic layout and containers it is quite suitable
17:36 sapier it's implementation, to be more precise the positioning is, is inconsistent and broken
17:37 sapier and for containers and dynamic layout we'd need a designer tool anyway
17:38 celeron55 voxus looks professional, but i'm yet to see any actually interesting content
17:38 sapier those guys seem to be skilled designers true
17:41 sapier I wonder if that guy suggesting the hw scaling did try it
17:42 sapier either I'm doing something different or irrlicht support for this is quite broken
17:42 Ritchie joined #minetest-dev
17:43 celeron55 and yes, i agree that the basic concept of formspecs is fine but 1) limiting, and 2) there are way too many historical problems
17:43 sapier I fully agree to point 2 ;-)
17:44 sapier as of code fixing it would be simple but I'm almost sure it'd cause major political issues
17:44 ImQ009 joined #minetest-dev
17:44 celeron55 did Zeno say he has experience of making UI systems?
17:44 sapier yes he did
17:45 celeron55 i wonder what kind of exactly
17:45 sapier but he didn't wanna say anything about him doing the formspec redesign
17:47 sapier ok seems like the irrlicht issues I just mentioned are due to changeing those parameters after creation of context .. guess I have to find a way to make them parameters of device creation
17:50 sapier nice latest irrlicht version contains support for reading assets
17:50 sapier maybe we can skip copying our textures to sdcard soon
17:52 celeron55 gah, i hate thinking about this
17:52 celeron55 everything is an absolute mountain of work and people's expectations are too high
17:55 celeron55 doing anything is not rewarding at all
17:55 sapier come on celeron55 everything is fine ;-)
17:55 sapier almost
17:56 celeron55 no, almost nothing is, and i can't help it
17:56 celeron55 well the world isn't ending, but otherwise
17:56 sapier crap I just tried to fix my system irrlicht header instead of the one within android deps :-)
17:58 celeron55 one day i tried to imagine how to mix up minetest and buildat in a useful way
17:58 celeron55 the end result was that it makes no sense and would be absolutely awful to make work in the first place
17:59 sapier as often
17:59 celeron55 yeah it's generally where these thought experiments go
17:59 celeron55 directly in the trash
18:00 celeron55 an another thing is that i would like to replace irrlicht with urho3d in minetest
18:01 VanessaE O_o
18:01 celeron55 and also take into use buildat's client-side lua wrapper of urho3d's gui system
18:01 sapier wouldn't this just exchange the problems we know about by problems we'd still have to find out?
18:01 celeron55 it's pretty functional i must say
18:02 celeron55 yeah, it would immediately bring in some issues like not being able to support very old GPUs
18:03 celeron55 would have to find out first how many users would be affected
18:03 celeron55 but it would solve a lot of issues too
18:03 sapier define "very old"?
18:04 celeron55 older than opengl 2
18:04 exio4 is there any gpu out there that can run minetest right now with that ogl version?
18:05 celeron55 of course it requires some specific extensions instead of that version so it's more complicated than that
18:06 sapier so basicaly anything below geforce 7
18:06 celeron55 EXT_framebuffer_object and EXT_packed_depth_stencil
18:06 celeron55 or something like that
18:06 sapier which is about 2005
18:07 sapier I'd guess intel required some more time to get opengl2 support
18:07 rubenwardy joined #minetest-dev
18:07 celeron55 my old laptop (the famous one that i made MT on) runs irrlicht on opengl 1.4
18:07 celeron55 so that's out
18:08 celeron55 also virtualbox doesn't support those extensions
18:09 celeron55 basically every 3D game has required those for ten years now though
18:09 celeron55 every proper one :P
18:09 sapier well intel wasn't as lazy as I expected their first ogl2 is 2005 too
18:10 sapier basically we're talking about limiting minetest to hw newer then 10 years
18:10 sapier with even some newer hardware not working
18:10 sapier does urho3d support android?
18:10 Calinou would Urho3D support shaders as well as Irrlicht currently does?
18:11 celeron55 sapier: it does
18:11 celeron55 Calinou: it's unable to operate without shaders
18:11 sapier is it official or would we have to do as much as we'd have to do for irrlicht?
18:11 Calinou ah, that's OK
18:12 celeron55 sapier: it officially supports android
18:13 sapier ok so by now no "no-go" issue ... what do we get if we switch?
18:13 celeron55 its build system probably needs to be reworked for minetest though, and it's not available as a distro package anywhere, and the author does not recommend anyone to package it
18:13 Calinou that's a rather big problem
18:13 celeron55 it's a bit of a binary focused build system
18:14 sapier that's nothing for the positive list celeron55 ;-)
18:14 celeron55 because it's made for traditionally distributed binary games and not for open source projects
18:14 celeron55 (but it's MIT licensed itself)
18:15 sapier I guess estimating switching the engine would cost about half a year to a year ain't overestimating
18:15 celeron55 sapier: well... we for example get a Lua interface to every part of urho3d
18:15 sapier we'd need quite big improvements to justify doing this
18:15 celeron55 which may or may not be helpful
18:16 sapier I'm not sure if this would be a benefit, it'd be way more we'd have to secure ;-)
18:16 celeron55 that's the problem, there will never be anything that can be just plugged in place of irrlicht and which happens to do exactly what we want
18:17 celeron55 i mean, i haven't properly gone through this evaluation myself either
18:17 sapier nope don't get me wrong that's not what I expect. No matter if it's irrlicht or any other engine we had right now switching it is a lot of work. It's only reasonable if the new engine is way more better
18:17 celeron55 it would probably be quite a cultural shock to those used to how minetest builds
18:18 sapier e.g. if the new engine would provide whole gui including description language, that'd make me think about a switch
18:18 est31 why not use html
18:18 est31 instead of formspecs
18:18 sapier html doesn't specify the "active" part
18:19 est31 no javascript
18:19 rubenwardy joined #minetest-dev
18:19 est31 but you can send forms
18:19 est31 and have buttons
18:19 celeron55 sapier: it provides a fully lua-controllable gui, which i use in buildat (the project that i have abandoned due to too much work) by sending the client some lua and running it in a sandbox
18:19 est31 and you have 100% finished rendering engines
18:19 sapier yes that's what makes html at least somehow fit to our current design
18:19 celeron55 the gui has some dynamic layouting features
18:19 est31 html has css
18:19 est31 (and other layouting features too)
18:20 sapier to be honest I don't wanna programm a gui in lua manually ;-)
18:20 Calinou libRocket eats tons of CPU
18:20 sapier especally if it's dynamic layout
18:20 sapier things like that require a designer tool
18:22 celeron55 http://fpaste.org/168382/00535142/
18:22 sapier but gui was just an example there could be other benefits worth it too
18:22 celeron55 here's some gui code in an experimental thing running on buildat
18:22 celeron55 you probably guessed it looks like that
18:22 celeron55 it's pretty much what an OOP gui system will always look like
18:23 sapier yes but still you can write simple guis that way but nothing complex
18:23 sapier it's like programming qt manually ... yes it is possible yet noone does it
18:23 sapier except some crazy ones maybe
18:24 celeron55 well, qt has QML which can only be programmed manually
18:24 sfan5 <VanessaE> here's the one that happened the other day on one of my servers:  http://digitalaudioconcepts.com/vanessa/h​obbies/minetest/screenshots/Screenshot%20​-%2001102015%20-%2001%3a03%3a25%20AM.png
18:24 celeron55 but it's more of a formspec-flavoured thing
18:24 sapier qtdesigner?
18:24 sfan5 it's a piece of art
18:24 celeron55 QML is a separate language
18:24 VanessaE sfan5: this by you is art? heh
18:24 celeron55 i think qtdesigner is only for reqular qt, at least last time i checked it was
18:25 sapier yes but for what I remember qml was added recently I'm quite sure they'll add a designer tool soon
18:25 celeron55 well yeah, it might be that they will add it
18:25 celeron55 anyway, if we want a gui that isn't programmed in that way, then we have no other option than to write it ourselves or just clean up formspecs a bit
18:26 celeron55 or a lot
18:26 sapier http://doc.qt.digia.com/qtcreator-2.3​/creator-using-qt-quick-designer.html
18:26 sapier I'm not against a gui programmed that way I just wanna have a designer for it too ;-)
18:27 sapier we could rename formspec to mtfl ... and fix the historic issues ;-)
18:28 celeron55 well, urho3d supports xml too, but irrlicht does that too
18:28 celeron55 i actually originally attempted to make some menus in minetest in irrlicht xml (before Lua); it was awful
18:28 celeron55 you don't get complete control with it
18:29 sapier Are you sure we'd get full control by urho3d?
18:29 celeron55 nope
18:29 celeron55 i'm not advocating urho3d's xml at all
18:30 sapier At best we'd use something exisiting. html(subset) qml the gtk ui stuff (don't know how they call it)
18:30 celeron55 things always get awful at the point when you have to center a menu on the screen and keep it centered when the window is resized and then if you want to make some HUD with it, god forbid
18:30 sapier but thats quite a lot of work
18:31 sapier right now I'd be fine with irrlicht findig the correct ui element for my touch events ;-)
18:31 sfan5 why would we even use html?
18:31 est31 joined #minetest-dev
18:31 celeron55 could we just remove UI
18:31 sfan5 and who'll write the html parser (hint: html is not xml)
18:31 celeron55 just make all menus as nodes in the game
18:32 celeron55 the main menu could be a world too!
18:32 sapier I'd love to see this celeron55
18:33 sfan5 hm
18:33 sfan5 mh
18:33 est31 sfan5 others. We just need to figure out how to get a renderer display content in front of ours
18:34 est31 there are a couple of permissively licensed html renderers
18:34 sfan5 random idea: you could make the redis backend connect to a remote redis server (that is read-only) and have like a continously updated tutorial world that way [or you could just do thing the normal way]
18:34 celeron55 html is not a gui language, it's awful doing that job
18:34 sfan5 est31: do you want to make minetest depend on webkit?
18:34 est31 um yea
18:35 sapier html is a form language at best
18:35 celeron55 conclusion: UI is insane, i don't understand why anyone would want to touch it
18:36 rubenwardy Doing the UI would interest me, but Minetest's code is insane.
18:36 rubenwardy Especially formspecs etc
18:37 celeron55 also i will respect sapier's insanity a lot if he still is willing to do *anything* to even design anything related to formspecs, lol
18:38 celeron55 i'm going to just live with what we have, it's absolutely not worth the effort to touch it
18:38 sapier well I'm gonna step back a few steps for some time ... I've been to close to quiting completely yesterday
18:38 sapier at least step back from formspec topic
18:39 celeron55 do you still happen to remember that the current networking implementation is a temporary solution until it is decided whether enet is good enough? 8)
18:41 * celeron55 imagines sapier is running away screaming at about 500 meters from his computer by now
18:41 sapier I know ... yet nobody ever did even try what I provided neither tcp nor enet implementation
18:42 sapier well at least the fixed version I had so much trouble to get in seems to work quite stable
18:42 celeron55 the whole thing was a stupid mess like usually
18:42 celeron55 but it seems by now that freeminer has proven that enet works well
18:42 rubenwardy It would be good to be able to do   size[width()/2;height()/2] or something like that
18:43 sapier Well I don't doubt enet works
18:43 Calinou Freeminer has enet, how well does it work there?
18:43 Calinou does it really provide lower ping?
18:43 sfan5 who rewrites the server ping code in the server list when we switch to enet?
18:44 sapier I don't think it's providing less latency
18:44 celeron55 how do we test this
18:45 sapier I don't know I never was able to benchmarks enet's latency or jitter as it just lies about it
18:45 celeron55 what's the general case where people complain of lag?
18:45 sapier at least it did
18:45 celeron55 a busy server hosted with a low-bandwidth connection?
18:45 sapier a jitter of 0ms is just impossible
18:45 sfan5 sapier: depends on the connection
18:46 sapier not even ping manages to get exactly 0.0 ;-)
18:46 proller joined #minetest-dev
18:50 sapier If you guys like to test it I'm gonna rebase the enet code, it's most likely not a big deal as whole network stack is built for replacing the low level protocol
18:50 proller joined #minetest-dev
18:51 celeron55 last time you said something in networking was easy you used weeks to create the current monster implementation 8)
18:51 sapier nope I built tcp and enet in parallel same time
18:52 celeron55 well if it's easy, go ahead
18:52 celeron55 then remind us here every day that we need a way to test it
18:52 VanessaE will the existence of it break old clients?>
18:52 sapier formspec android and network protocol same time? are you sure you wanna be that insane
18:53 sapier If we do it the way I implemented enet code it's gonna be possible to run both protocols in parallel
18:54 sapier at different ports of course
18:54 celeron55 well we will certainly drop the old protocol after a version that supports both
18:54 sapier that's how I intended to do the transition
18:54 VanessaE that's fine
18:54 VanessaE I only ask because...well...you said you need testers/.
18:55 VanessaE and I've got several production servers that could be users as long as old clients can still play normally.
18:55 celeron55 yes, that seems somewhat useful for testing, altough i wonder how much the bandwidth competition between the protocols affects the result
18:55 VanessaE used*
18:56 sapier well you shouldn't mix it on testing ;-)
18:59 VanessaE well one of those servers is expressly intended for testing.
19:00 VanessaE I can move it to the enet port (or whatever) and let you all go at it there
19:00 VanessaE it's not heavily modded, just technic and plantlife (and they're mostly already played out) so it'll respond as fast as any other server should.
19:00 sapier yet it's not really usefull to compare enet to our current protocol. Anything but enet beeing better would be a surprise
19:01 sapier comparing enet to tcp would be way more open ;-)
19:12 Calinou TCP for real-time games?
19:12 Calinou enet was written because TCP was slow…
19:12 sapier Calinou: do you really think enets "reliable" communication is faster then tcp? ;-)
19:13 Calinou nope
19:13 Calinou but TCP makes everything reliable (AFAIK), which maybe isn't what you want…
19:13 Calinou is TCP programming easier or harder than UDP?
19:13 Calinou switching to TCP may simplify the code, if that is the case
19:14 sapier well if we did use tcp we'd have to use a mixed mode protocol
19:14 sapier partial udp partial txp
19:14 sapier tcp
19:15 sapier about as complex as udp, but way more fast if you wanna transmit large data
19:15 sapier e.g. textures
19:15 Calinou ah, right
19:15 Calinou if you do that, be sure to make an announcement, so that people update their port forwardings
19:15 MinerDad joined #minetest-dev
19:15 sapier if I remember correct it's been about factor 100 to current and still about factor 10 compared to enet
19:16 sapier no 100 was compared to old implementation current is quite a lot faster ... yet not capable to beat enet
19:29 celeron55 i think enet would be a good compromise between simplicity and performance
19:30 celeron55 it's going to max out the general internet connection anyway
19:31 celeron55 or, i mean, average
19:34 johnnyjoy left #minetest-dev
19:34 johnnyjoy1 joined #minetest-dev
19:35 VanessaE celeron55: define "average"
19:35 VanessaE here, that could mean 5 Mbps (sadly)
19:36 celeron55 dunno, probably under 1MB/s
19:36 Calinou this would make remote server more redundant
19:36 est31 64 kb is enough :D
19:36 Calinou if we can reach 1 MB/s
19:36 jluc joined #minetest-dev
19:37 celeron55 i think the goal should be making it redundant
19:37 celeron55 so let's make sure to test that as one of the first things
19:41 Calinou yeah, it's good not to need a Web server
19:41 Calinou initially that was an hack
19:41 johnnyjoy1 left #minetest-dev
19:42 sapier Calinou: right now webserver is even slower then directly downloading textures
19:45 PilzAdam joined #minetest-dev
19:59 crazyR joined #minetest-dev
20:07 PilzAdam sapier, is it intentional that the mainmenu now scales with the window size?
20:09 sfan5 PilzAdam: probably intentional (still looks bad IMO)
20:10 VanessaE intentional, but a patch was proposed already
20:11 VanessaE https://gist.github.com/sa​pier/846662f8f8aecaff166a
20:11 VanessaE this, I think it was.
20:11 sapier true
20:15 rubenwardy http://rubenwardy.github.io/mine​test_doc/chapters/releasing.html
20:15 rubenwardy oops
20:17 Calinou “protection”
20:17 Calinou copyright is not protection
20:17 Calinou it's a vicious term :/
20:18 Calinou just my two 0.02 €
20:18 Calinou wrong channel anyway
20:19 sapier damn
20:19 sapier I never ever experienced some code just working after rebase
20:22 sapier celeron55 https://github.com/sapier/mine​test/tree/network_addon_enet_6 ;-) rebase completed
20:23 crazyR joined #minetest-dev
20:23 sapier don't expect it to be polished there are a few parts missing yet everything required for testing is in there
20:27 acerspyro joined #minetest-dev
20:29 sapier #2111 ... no comment
20:29 ShadowBot https://github.com/minetest/minetest/issues/2111 -- Problems with building
20:34 sapier kilbith: fixed or still open ? https://github.com/minetest/minetest/issues/2077
20:34 ezraanderson joined #minetest-dev
20:35 kilbith regarding the general complaining actually, still open.
20:35 sapier so formspec is still broken?
20:35 ezraanderson +sapier, do me input positions, x & y are in the wrong spot?
20:36 sapier ezraanderson: yes that's what seems to happen if I press somewhere the gui acts somewhere else
20:36 kilbith sapier, not broken,  that's not the subject of the issue
20:36 ezraanderson give me 5, minutes, I have never tried it on an android phone
20:37 sapier sorry for asking kilbith I'm already getting headache again ... can you please settle this with Zeno and vanessae?
20:37 sapier ah and sfan5 of course
20:37 kilbith could you just try the suggestion of VanessaE yesterday ? (pixels = points/72*DPI)
20:37 sapier nope
20:38 sapier to get this done I'd have to touch almost any part of formspec
20:38 OldCoder joined #minetest-dev
20:38 sapier result would be another variant
20:39 kilbith so you consider the current state as your best result you can do ?
20:39 kilbith (honestly)
20:39 sapier especially as I don't even exactly understand what pixels is meant because points/72*dpi will return some float value
20:40 sapier I already said two days ago I'm not gonna tune anything at current code the only further work for formspec I have provided was reverting back to 0.4.9 style ... which wasn't accepted too
20:41 sapier if I interpret vanessae's suggestion correct it's already implemented that way
20:41 kilbith we'll discuss further with Zeno to see how things can evolve
20:41 MinetestForFun joined #minetest-dev
20:42 sapier do it this way
20:45 kilbith well, actually it's no longer really an issue (chat bug, header violation by tabs etc.) nor a good result either
20:45 kilbith so i guess i can close it
20:46 sapier define "good result"?
20:47 kilbith oh no, not again :'(
20:47 sapier well you're using those subjective terms ;-P
20:47 acerspyro Could be the politics "good result" expression, too.
20:48 kilbith my humble suggestions actually is to define 0.85 as default scaling and make the fonts a little more fat
20:49 kilbith but yeah, Irrlicht sucks for that
20:49 sapier well I don't care about forst one
20:49 sapier but I thought I'd have made clear we can't make this font more fat we'd have to exchange the font
20:51 kilbith initially, the user don't have to bother with .conf or strafe the settings tabs of click for adjust the scaling
20:51 kilbith so you should consider 0.85 as default
20:52 VanessaE kilbith: just change your default font face.
20:52 VanessaE font_path = /home/vanessa/.minetest/fonts/DejaVuSans.ttf
20:52 VanessaE mono_font_path = /home/vanessa/.minetest/fonts/DejaVuSansMono.ttf
20:52 VanessaE in minetest.conf
20:52 VanessaE point them to fonts that look good for you
20:52 kilbith i normally don't have to bother changing the font face
20:53 acerspyro Setting up the game should be easier than this.
20:53 acerspyro I agree.
20:53 sapier suggestions please not agreements ;-)
20:53 kilbith that ain't user-friendly
20:54 kilbith and a good game normally is for relax
20:55 sapier hmm we could sell mintest bundled to a pc including screen this way we could be sure it looks exaclty like it looks for us ;-)
20:55 acerspyro Dynamic font size?
20:55 sapier I know about a huge company doing it this way ... works quite good for them
20:56 proller joined #minetest-dev
20:56 acerspyro Strangely, I don't recall having to change the font size in any other game...
20:57 acerspyro I only play OSS stuff, btw.
20:57 sapier acerspyro: do you HAVE to do it here?
20:57 acerspyro I would love to
20:58 sapier well that's great because I can't remember any game where you even CAN change the font ;-)
20:58 acerspyro Cube 2 games
20:58 sapier I don't think there's alot of text in those games ;-)
20:58 acerspyro There's the chat, the death messages, the info messages...
20:59 sapier doesn't change the fact minetest already supports more features then most other games ;-)
21:00 acerspyro Wrong
21:00 acerspyro You can't even move your head.
21:01 sapier I don't remember head beeing related to fonts
21:01 acerspyro You changed the subject to the fact that MT supports more features than any other game
21:01 acerspyro While it can't do chat colors either, afaik
21:02 sapier I was talking about fonts only
21:02 sapier no need to switch topic as we always have been talking about fonts
21:13 kilbith i just tester the DejaVuSans font and the result is worse than before : https://lut.im/8Yv98312/sxTeGIFP
21:13 kilbith tested*
21:13 kilbith awful
21:14 kilbith unless if you're strongly myopic
21:15 acerspyro omg dem static menus
21:15 VanessaE kilbith: gui_scaling is set back to normal?
21:15 acerspyro y u no dynamic GUI?
21:15 kilbith yep, scale cleaned in .conf
21:15 kilbith at 1.0 in that screenie
21:15 VanessaE kilbith: then just turn the font size down.  obviously DPI isn't detected correctly on your machine
21:16 VanessaE font_size = 10
21:16 VanessaE mono_font_size = 10
21:16 VanessaE see what that does
21:17 VanessaE (just guessing based on your screenshot)
21:18 kilbith https://lut.im/oT1V4QFS/rEY80DIE
21:18 VanessaE ok I was close.  try 12 instead of 10
21:18 VanessaE better yet, 14
21:18 VanessaE I'm curious.
21:19 kilbith 14 : https://lut.im/dkgG99hQ/iXLrfP6h
21:20 VanessaE ok then for you I guess 12 or 13 would be about right.
21:20 kilbith and i repeat : i normally don't have to bother with .conf like that.  the average user won't.
21:20 acerspyro ^The words I couldn't find
21:20 VanessaE sure, but first we need to find out the difference between you and what "should" happen
21:20 acerspyro "The average user won't"
21:22 VanessaE (I wanna know why kilbith's fonts are coming up ~50% larger than they should be to begin with)
21:22 kilbith well, 13 is nice : https://lut.im/Gn68MNiK/sCQy8y8B
21:23 VanessaE kilbith: yeah, that's about consistent with what should normally be - and you're sure you're at git HEAD now?
21:23 sapier kilbith: are you on linux or windows?
21:23 acerspyro sapier: facepalm for you sie
21:23 Calinou he's not on Windows, look at window decorations
21:23 acerspyro sir*
21:24 VanessaE I can't begin to guess what distro or DE that even is..
21:24 acerspyro Unless LXDE has been ported to Windows...
21:24 kilbith VanessaE: was testing on latest launchpad build - but i can test with Git HEAD, it'd identical
21:24 sapier or he uses some lxde theme
21:24 acerspyro VanessaE: It's LXDE, looks like Mint
21:24 kilbith sapier: Lubuntu 14.04
21:24 acerspyro Or another Ubuntu variant
21:24 VanessaE kilbith: latest launchpad build?  how old is that anyway?
21:24 sapier is lubuntu still x11?
21:24 kilbith about 20h
21:24 acerspyro VanessaE: very.
21:25 acerspyro :P
21:25 Calinou yes, sapier
21:25 sapier 20h :-)
21:25 Calinou almost all distributions use Xorg by default
21:25 kilbith https://code.launchpad.net/~minetestdevs​/+archive/ubuntu/daily-builds/+packages
21:25 sapier well that's ancient
21:25 Calinou does Minetest run on Wayland or Mir (without compatibility layers) anyway?
21:25 kilbith 22h old to be exact
21:25 sapier Calinou: that's gonna change soon
21:25 acerspyro 0.4.11 came to OpenSUSE's repos about a day after its release.
21:26 kilbith sfan5 seems use the same default font as me too
21:26 sfan5 Calinou: i don't think it does
21:27 sapier 22h is quite useless to discuss about while we're about to fix it you meed latest HEAD to be able to participate the discussion
21:27 VanessaE and 22h ago, who even knows how old the commits are that went into it?
21:27 sapier Calinou: compatibility layers just provide compatibility not necessary same behaviour
21:27 kilbith always trying to invalidate my problems, eh ;)
21:27 VanessaE kilbith: basic tenet of science ;)
21:28 kilbith https://github.com/minetes​t/minetest/commits/master
21:28 VanessaE "Make me a liar, Fish!" (Dr. Ellie Arroway, "Contact")
21:28 kilbith ^ latest build built after b7c...
21:28 kilbith so your remark is irrelevant sapier
21:29 sapier if it's b7c0e4b3336ef5bdb0bdd6baa3257787886b2c73 your build should be fine to discuss the issue
21:30 kilbith the followed commits concerned android only
21:30 sapier then try to find out why it's looking that different for you
21:30 kilbith but ok, i'll compile right now
21:31 sapier I suggest debug build and RUN_IN_PLACE to have commit in screenshots as well as be sure no other minetest.conf file is used
21:31 kilbith sure
21:31 kilbith hmm, debug build ?
21:32 sapier -DCMAKE_BUILD_TYPE=Debug
21:32 Calinou sapier, thanks for increasing step height on Android!
21:32 sapier Calinou: I don't know why this wasn't done some time ago the suggestion was there for about half a year
21:52 kilbith what's the... https://lut.im/mcavuaCy/Mr01yp1u  first starting, not even a .conf generated, and the scaling is iniatially perfect O.o
21:52 kilbith Vanessa, sapier ^
21:53 sapier ok now build release to make sure it's not behaving different for you
21:53 sapier DCMAKE_BUILD_TYPE=Release
21:53 kilbith 'k
21:54 VanessaE ok then we know you had something bad in your conf
21:54 VanessaE (because I run a debug build)
21:54 sapier he does do this right now to
21:55 VanessaE oh right
21:55 VanessaE misread.  I need new glasses or eyes or something.
21:55 kilbith no, my previous Git build (2 days ago) with RUN_IN_PLACE=1 behaved like the launchpad build
21:55 sapier well I need a new keyboard it's always missing or exchanging keys
21:55 kilbith with a virgin .conf
21:55 sapier 2 days ago is quite old
21:59 VanessaE it's interesting to note that his locale reverted to english..
22:12 kilbith Release build, initial display : https://lut.im/LmYdSk31/KT1qKG0M
22:12 kilbith ^ sapier
22:13 VanessaE lgtm.
22:13 kilbith thats just similar
22:13 sapier looks like current head is fine?
22:14 kilbith yes, fine :)
22:14 sapier is it possible that dayly build uses some config file you missed?
22:14 kilbith so i take off what i said to you, it's a good work and i close my issue
22:15 VanessaE well wait
22:15 VanessaE sapier's question is valid
22:15 VanessaE what is the buildbot on launchpad doing to fuck up the fonts?
22:16 kilbith but there's still a zone of mystery : yesterday i removed my .conf before installing the latest .deb, this .deb has been built before the android commits only, and i still had those monstruous fonts...
22:16 kilbith X-files.mp3
22:18 VanessaE now who gets to merge sapier's fixed-size-formspecs gist? :)
22:20 MinerDad joined #minetest-dev
22:22 VanessaE sapier: I know you don't want to touch formspecs but imho the best way to solve the entire mess is a revamped formspec language (whatever form it takes - HTML, XML, formspec tables, who knows) but with a slightly different method of summoning them
22:23 VanessaE just like with two network stacks, the old way could be marked deprecated, read-only like you suggested, and be discontinued some day in the far future
22:23 VanessaE or link against QT ;)
22:26 sapier he just told the current variant is fine if he doesn't like the one after freverting those changes I'll have to do this discussion again. I provided the code. It's up to someone else to decide if it's supposed to be that way.
22:26 VanessaE sapier: right.  the fonts are good in 99% of cases now I think.
22:46 acerspyro The hand item should really be the player's arm with the skin on it, like in Minecraft. The potato is just really weird.
22:47 sapier mt != mc
22:48 ezraanderson :)
22:48 n4x joined #minetest-dev
22:51 ezraanderson mt > mc ?
22:52 acerspyro sapier: you really need to make a book on killing conversations you don't want to take place. You're really good at it .-.
22:59 sapier If I were good at it it'd not have taken 5 days to settle that dispute
23:01 VanessaE sapier: to be fair, most of that time was wasted due to time zone differences
23:01 VanessaE lemme rephrase that
23:02 VanessaE most of that wasted time*
23:02 sapier considering this dispute beein spread across three continents that's been one of the factors true
23:03 VanessaE I'd say it was the main factor
23:03 VanessaE it's hard enough to get two people to agree on anything here as it is :)
23:03 VanessaE (that's when they're already together)
23:04 VanessaE then we have kilbith's issue caused by launchpad's build being somehow botched
23:04 sapier I wouldn't say it's main factor but lets let it stay what it is ... history
23:05 VanessaE yeah
23:21 celeron55 i would love to see what a mailing list would do to these discussions
23:22 celeron55 i'm not a fan of them, but from a practical standpoint it would seem like it would encourage actually thinking things and not just shouting what comes to mind
23:25 celeron55 well, or, we could just try to move onto the forum
23:27 celeron55 dunno, i'll probably do nothing to achieve any of this; just a thought
23:54 acerspyro joined #minetest-dev
23:54 sapier left #minetest-dev

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