Minetest logo

IRC log for #minetest-dev, 2014-05-23

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

All times shown according to UTC.

Time Nick Message
00:08 Sokomine where can the official android port be found again? i downloaded the last ones as debug builds from sapier
00:13 VanessaE http://animalsmod.comuf.com/downlo​ads/Minetest-release_20140518.apk
00:13 VanessaE that's the latest one I could find
00:20 Sokomine thank you :-)
00:44 diemartin joined #minetest-dev
00:46 diemartin comments? https://github.com/kaeza/minetest/commit/​119589e36d44bf7555406d81262405a18a95422d
00:49 VanessaE <proller> 1000 useless lines commit
00:49 diemartin D:
00:57 cheapie joined #minetest-dev
01:01 cg72 joined #minetest-dev
01:10 cheapie joined #minetest-dev
01:12 cg72 cheapie your in the dev chat now???
01:12 cheapie I have had it set to auto-join for a while now.
01:13 cheapie I got my auto-login script working, finally...
01:13 cg72 im here but i dont think i help or will ever for that matter :(
01:14 cheapie After I join, I can just do "/py load login.py" and HexChat logs me into all 7 of VE's servers.
01:15 cg72 hey i need that script
01:15 cheapie http://pastebin.ubuntu.com/7503749/
01:17 cg72 what do i save it as and where?
01:17 cheapie Of course, HexChat thinks I'm being /msg flooded when I do that, and disables tab auto-opening. It still logs me in, though.
01:17 cheapie I saved it as login.py in my home folder. That way, I don't have to specify the path, just the name.
01:18 cg72 thanks
01:21 Taoki joined #minetest-dev
01:32 Taoki joined #minetest-dev
01:50 Megaf_ joined #minetest-dev
01:52 Bhima__ joined #minetest-dev
01:59 ShadowNinja kaeza: Commented.
01:59 cheapie joined #minetest-dev
02:06 Taoki_1 joined #minetest-dev
02:10 Taoki joined #minetest-dev
02:40 kaeza ShadowNinja, oops. fixed indentation
02:40 kaeza re: interact_extra, I don't see it documented anywhere, shall I remove?
02:43 kaeza grep says it's only used on that very line...
02:47 BrandonReese joined #minetest-dev
02:58 RealBadAngel joined #minetest-dev
03:11 ShadowNinja kaeza: Yes, that's the only place I've seen it.  Probably a legacy priv that wasn't completely removed.
03:16 zat Any ideas on how to be able to add a lua function to the list of decorations in register_decoration?
03:16 zat So some decorations can be generated by Lua code.
03:18 zat I tried to do it by keeping the lua_State* and a reference to the function in DecoSimple, then I called it with the coordinates of the decoration as argument.
03:18 zat ...the result was crappy, partial trees and segfault after a few seconds.
03:19 zat (Code available if someone wants to see)
03:37 Exio4 joined #minetest-dev
03:40 cheapie joined #minetest-dev
05:40 OdinAllfather joined #minetest-dev
05:41 Hunterz joined #minetest-dev
05:41 OdinAllfather left #minetest-dev
06:05 darkrose joined #minetest-dev
06:29 Bhima joined #minetest-dev
06:37 jin_xi joined #minetest-dev
07:23 Bhima_ joined #minetest-dev
07:25 vifino joined #minetest-dev
07:26 vifino joined #minetest-dev
07:27 vifino joined #minetest-dev
07:46 kaeza joined #minetest-dev
07:57 kaeza hmm
07:57 VanessaE ?
08:01 kaeza any specific reason for this? https://github.com/minetest/minetest/blo​b/master/builtin/game/privileges.lua#L10
08:02 VanessaE I'd guess just old code from when someone was still learning Lua?
08:04 kaeza guess so
08:04 * kaeza fixes
08:11 Bhima___ joined #minetest-dev
08:17 kaeza ShadowNinja, pull updated
08:39 grrk-bzzt joined #minetest-dev
08:51 PilzAdam joined #minetest-dev
09:10 grrk-bzzt joined #minetest-dev
09:35 ImQ009 joined #minetest-dev
10:12 proller joined #minetest-dev
10:19 restcoser joined #minetest-dev
11:12 vifino joined #minetest-dev
11:22 kahrl joined #minetest-dev
11:32 sfan5 pushing http://sprunge.us/UBSC?diff in 5 mins
11:43 Megaf joined #minetest-dev
11:44 PenguinDad joined #minetest-dev
11:54 bookwar1 joined #minetest-dev
12:00 Megaf_ joined #minetest-dev
12:05 deltib joined #minetest-dev
12:07 proller joined #minetest-dev
12:19 deltib joined #minetest-dev
12:44 deltib joined #minetest-dev
12:49 Zeitgeist_ joined #minetest-dev
12:49 Zeitgeist_ joined #minetest-dev
12:56 proller joined #minetest-dev
13:08 proller joined #minetest-dev
13:18 deltib joined #minetest-dev
13:18 smoke_fumus joined #minetest-dev
13:27 proller joined #minetest-dev
13:43 vifino joined #minetest-dev
13:44 vifino joined #minetest-dev
14:15 hmmmm joined #minetest-dev
14:31 RealBadAngel ok, since #1308 has already 3 votes
14:31 ShadowBot https://github.com/minetest/minetest/issues/1308 -- Item entity stacks merge on the ground, add TTL. by RealBadAngel
14:32 RealBadAngel any objections or i merge it
14:33 RealBadAngel but please no "why do u hate items - proller's style"
14:33 kaeza lol
14:39 proller increase limit
14:40 RealBadAngel what limit you mean?
14:41 RealBadAngel ttl?
14:41 proller yes
14:42 RealBadAngel no problemo, i will add if player:get_player_name()="proller" then item_entity_ttl = forever end
14:43 RealBadAngel c'mon, grow up man
14:43 Amaz ^ +1
14:44 * sfan5 meows
14:45 sfan5 any comments on https://github.com/minetest/minetest/pull/1149 ?
14:45 sfan5 ~title
14:45 ShadowBot Add AppData file by davidgumberg · Pull Request #1149 · minetest/minetest · GitHub
14:56 Jordach joined #minetest-dev
15:04 kaeza any comments on #1324 ? (ignore the second commit pls)
15:04 ShadowBot https://github.com/minetest/minetest/issues/1324 -- Sort commands and privs alphabetically in `/help. by kaeza
15:06 RealBadAngel ok, since sapier, SN and VE are for, im mergeing 1308 now
15:08 RealBadAngel sorry, proller, another rebase nedeed
15:11 sfan5 kaeza: looks good
15:27 CraigyDavi joined #minetest-dev
15:29 Megaf joined #minetest-dev
15:44 Jordach joined #minetest-dev
15:46 Hunterz joined #minetest-dev
15:49 Zeitgeist_ joined #minetest-dev
15:49 Zeitgeist_ joined #minetest-dev
15:52 sapier joined #minetest-dev
15:52 crazyR joined #minetest-dev
16:12 rubenwardy joined #minetest-dev
16:21 restcoser joined #minetest-dev
16:33 zat joined #minetest-dev
16:49 Megaf That builtin_items mod is using a lot of CPU
16:49 Megaf sapier; any idea how do I slow it down?
16:50 sapier builtin_items?
16:50 Megaf yep
16:50 Megaf you said it was checking for changed every X us
16:51 sapier we're talking about the changes rba did recently or something else?
16:53 sapier ok that one isn't merged so no
16:53 sapier can you show me what we're talking about?
16:54 Megaf no idea
16:55 sapier the only way to reduce mod cpu usage is increase server step
16:55 Megaf anyway, this shit is again using 99.9% of a single core and lagging like hell
16:55 sapier at least if the mod isn't completely broken
16:55 sapier we're talking about your pi are we?
16:55 Megaf no...
16:55 sapier but?
16:56 Megaf stop blaming my hardware
16:56 Megaf minetest is terribly coded
16:56 sapier nope
16:56 Megaf almost no use of threads and no use of SMP at all
16:56 sapier it's designed to use as much cpu as possible
16:56 Megaf maybe a little on mapgen
16:56 Megaf # Length of a server tick and the interval at which objects are generally updated over network
16:56 Megaf dedicated_server_step =
16:56 sapier the only way to reduce server thread cpu usage is increase the server step
16:56 Megaf is that in seconds?
16:56 sapier yes
16:57 Megaf how does that works?
16:57 VanessaE no, he uses an odroid.  :P
16:57 sapier server will sleep for that time, meaning this will be the minimum reaction time for any user interaction
16:57 Megaf how can that give a better performance?
16:57 VanessaE and if your server is lagging it's because either your mapgen is finally kicking in (new terrain) or you screwed up that network parameter
16:58 sapier hmm I'd have to check but It could be a minimum value too thus if your cpu isn't fast enough to handle all operations within that time it's gonna run at 100% load
16:59 Megaf well, changed that to 1 instead of 0.1, now ServerThread uses at most 15% of CPU
16:59 sapier yes and any reaction will take at least 1s + rtt
17:00 VanessaE and players' movements will be incredibly laggy too
17:00 Megaf and it jumped to 99.9% when I did a simple /status
17:00 sapier if your cpu can't handle the operations that need to be done per step there's no parameter you can change to fix it
17:00 sapier it's like trying to win a f1 race with a lawn mower
17:01 Megaf Can't you just optmize the code?
17:01 sapier no
17:01 VanessaE can't you just use a piece of hardware that's designed for server use?
17:01 VanessaE odroid is not designed for this
17:01 Megaf stop creating new features
17:01 sapier no matter how much you optimize there's till a lower boundary for cpu requirement
17:01 Megaf and begin correcting bugs and reducing the code
17:01 proller it was already optimized in freeminer ;)
17:01 sapier and effort to win each % down there increases exponentially
17:01 Megaf celeron55 made minetest to run on low end hardware
17:02 Megaf thats the objective of minetest
17:02 proller no.
17:02 VanessaE Megaf: low-end as in a low-end intel machine,
17:02 sapier non x86 isn't low but ultralow hardware
17:02 proller low-end core i5
17:02 VanessaE not a low-end ARM with a shitty CPU-bound network device
17:02 Megaf man, you are all going backwards
17:02 Megaf world is moving towars efficiency
17:02 sapier you need about factor 10 mhz to achieve the speed of x86 for most operations
17:02 Megaf you need a fast single core CPU
17:02 Megaf common
17:03 Megaf go and take some computer science class
17:03 sapier no you don't megaf
17:03 * Megaf attended to Computer Engineering
17:03 sapier we're paralellizing minetest more and more but that can't be done in no time
17:04 VanessaE and it won't fix CPU-bound hardware like your crappy network device
17:04 sapier spliting mods from server step is the next step
17:04 proller sapier, yea, like split connection to two threads!
17:04 Megaf VanessaE; network hardware is no longer causing any problem here
17:04 proller need more threads!
17:04 VanessaE Megaf: then why does the CPU react so suddenly to network activity?
17:05 Megaf proller; shut up, freeminer is shit too
17:05 Megaf VanessaE; to load something from disk
17:05 proller Megaf, then buy minecraft
17:05 VanessaE Megaf: wrong.
17:05 Megaf proller; I did, long ago
17:05 proller buy again
17:05 Megaf I have plenty of power to use minecraft
17:05 VanessaE Megaf: explain the reaction to the /status  command
17:06 sapier the only way to have a cance to get minetest fly on devices with slow but a lot of cpu's is split it ... but that's the backside, the more cpu's the more overhead and the more to do wrong on locking
17:06 Megaf I really like minetest mroe
17:06 Megaf more*
17:06 Megaf just really frustrated too
17:06 sapier meaning we will have bugs as we do it
17:06 sapier and we wont get the most optimal solution from the beginning
17:07 Megaf maybe you should just start from scratch
17:07 sapier I suggest trying a little bit and find a server step causing about 60% cpu load on your device, that's not perfect but may result in acceptable balance between reaction time and cpu load
17:09 * VanessaE gives up
17:10 celeron55 on what hardware is Megaf running the server?
17:11 Megaf celeron55; 4 core ARMv7, 2,1 GHz
17:11 Megaf 2GB DDR2
17:11 Megaf Minetest will never use more than a single core
17:11 Megaf and ServerThread will alone use 99.9% of a single core
17:12 sfan5 ServerThread could be split up into serveral threads
17:12 sfan5 I agree on that
17:12 Megaf I can't even see what exactely is using the CPU, if its a mod, world loading, or anything
17:14 sapier starting from scratch wouldn't solve the main issue
17:14 VanessaE celeron55: an odroid, to be specific.
17:14 VanessaE USB-interfaced network hardware
17:14 sapier for what I know none of us did ever write a full parallel game so we need to learn a lot of things until we're done
17:15 sapier 4 core arm 2.1GHZ ... hmm 4 core 200mhz pentium
17:15 sapier that's gonna be tough to optimize that far
17:16 celeron55 well, really all i can say is that it isn't really designed to use multiple cores; it uses multiple cores where it reasonably can (which is basically only non-modded world generation)
17:17 celeron55 sapier: each core is more like a pentium 4
17:17 celeron55 in terms of performance
17:18 sapier depends on what operations you do
17:18 celeron55 which is... well, not much
17:18 celeron55 (a high-end pentium 4)
17:19 sapier I don't have a p4 so I can't test ;-)
17:19 celeron55 my server is a p4
17:19 celeron55 there is a reason why i don't want to run a minetest server on it anymore
17:19 celeron55 the reason is not hard to guess
17:19 sapier p4 with hyperthreading?
17:20 celeron55 yes, it's quite a high end model
17:21 sapier I'm not sure if we ever achieve to run a minetest server fluently on odroid but I assume spliting mod step from server step could help ... total system load might increase but (hopefully) spread to different cores
17:21 sapier at least if we get lockig right
17:21 celeron55 i would guess most serious servers run on single-core VPSes so i'm not sure if better multithreading should be considered a short-term priority
17:21 celeron55 it's probably a terrible amount of work to get anything reasonable to be done on a new thread
17:22 sapier optimizing to both single and multithreading is almost impossible
17:23 celeron55 at least with what we have, yes
17:23 sapier and atm I believe that future is more like multiple cheap cores instead of 1 big fast core ... but of course that's just an assumption
17:23 celeron55 there are modern ways to design software so that it can scale well to any number of cores but that doesn't really apply to minetest
17:25 Calinou joined #minetest-dev
17:25 celeron55 nor really generally to anything
17:25 sapier in theory you're right, but in practice I've never seen this to work for "small" workloads
17:25 sapier and while minetest is too much for odroid it's still very very small
17:25 celeron55 well, world generation is a task that is very threadable
17:25 celeron55 but that's why it already is
17:26 celeron55 the problem is that often lua is used instead, which isn't threaded
17:26 sapier yes and spliting user interaction from mod handling could improve "felt" performance too
17:26 celeron55 (...and then there's lighting)
17:26 celeron55 (and certain other nonlinear things)
17:27 sapier acyclic and cyclic things
17:27 sapier by now both are handled in same thread
17:30 sapier but prior starting another big task I'm gonna complete the ones I already started. first thing android
17:32 proller you never complete your tasks
17:34 Megaf sapier: VanessaE: celeron55: my minetest.conf http://paste.debian.net/101453/
17:35 Megaf I just changed it a little, lets see what happens
17:36 Megaf server step is set to 0.4
17:38 nore joined #minetest-dev
17:38 Megaf !up minetest.megaf.info 30003
17:38 ShadowBot Megaf: minetest.megaf.info port 30003 is up (620ms)
17:40 sapier what cpu load do you get for 0.4?
17:40 BlockMen joined #minetest-dev
17:40 Megaf 24% with no one oline
17:41 VanessaE inb4 "20%---no wait 99%"
17:41 VanessaE Megaf: how is the storage media interfaced?
17:42 VanessaE PCI or similar, or USB?
17:42 Calinou <Megaf> server step is set to 0.4
17:42 Megaf let me check
17:42 Calinou I would like to see moving players
17:42 Calinou :phoronix rolleyes:
17:42 Calinou 0.2 makes them sluggish already
17:42 Calinou (this is why it was decreased to 0.1)
17:43 * VanessaE bets on USB for storage also
17:43 Megaf fuck
17:44 Megaf it is nto
17:44 Megaf not*
17:44 VanessaE what is it then?
17:44 VanessaE something worse? :P
17:44 Megaf the bus is MMC I think
17:45 Megaf MMC 2 bit
17:45 Megaf ops
17:45 Megaf MMC 2, 4 bit
17:45 sapier come on guys stop punching at megaf, hardcore tests are the most valuable ones. Even if his setup is hard to get to fly it's providing quite interesting information. Any improvement we achieve there will help everyone else too
17:46 VanessaE Megaf: and how is that MMC bus interfaces to the rest of the system?
17:46 VanessaE sapier: not punching at megaf, punching at the device he's abusing
17:46 Megaf Maybe I should just give up and do like a lot of minetest admins are doing, going to micecraft servers
17:46 VanessaE Megaf: answer the question
17:46 VanessaE I want to know if this bus is bitbanged or properly interfaced.
17:46 Megaf storate > MMC > CPU
17:47 Megaf just like any SSD powered laptop
17:47 sapier knowing minecraft does really handle this better would be an interesting information too
17:47 VanessaE unless I've missed something, you can't just do MMC -> CPU.  something has to be between them to handle the hardware interfacing.
17:49 VanessaE (of course odroid uses a SoC device so things work differently there, but to the CPU the same rules still apply)
17:50 VanessaE and SSDs, Megaf, are usually SATA-interfaced, which means they go through a chipset designed to interface that bus to the CPU and its bus
17:50 Calinou Megaf, no, you can plug a SSD in SATA, mSATA, miniPCI-E or m.2
17:50 Megaf http://dn.odroid.com/homeba​ckup/201312201654587307.jpg
17:50 VanessaE i.e. the so-called "chipset"
17:50 VanessaE Calinou: ninja'd :)
17:50 Megaf sapier: this board can handle a full featured Debian Desktop, run libre office, firefox, HD videos and so on
17:51 VanessaE Megaf: that tells us exactly zero
17:51 VanessaE HOW IS IT CONTROLLED?
17:51 * sfan5 silently lols
17:51 celeron55 i don't think the interface matters much at all; what matters is that it's a small flash card which means it's very slow
17:51 celeron55 and especially slow at writing things
17:52 sapier how much ram is there?
17:53 Megaf celeron55: 11 MB/s right
17:53 Megaf up to 30 MB/s read
17:53 VanessaE celeron55: well it's precisely that, but the distinction I was trying to make was whether the CPU has to do all the bitbanging work in writing data into the card, versus at least some hardware assistance with memory mapping et al.
17:53 sapier if there's enough to create a ramdisk for a small world you could try this to check if it's really an io limitation
17:53 celeron55 Megaf: so 5 times slower than a HDD 8)
17:53 celeron55 or 10
17:53 Megaf It's a class 10 SD card, and no, its not the SD Cards fault
17:53 Megaf celeron55: do some random writes on your disk to see how fast it really is
17:54 proller but freeminer works good on first, slower odroid, and acceptable on raspberry pi..
17:54 sapier megaf I don't think theres "the issue" it's a lot of things
17:54 proller maybe its not hardware problem ?
17:54 sfan5 proller: could you actually be helpful instead of always saying "it works better in FM"
17:54 * Megaf will slowly count to ten
17:54 proller sfan5, yo can look on diff ;)
17:54 Megaf freeminer wont even compile here, so shut the fuck up proller
17:55 sfan5 ^
17:55 * sfan5 silently laughs
17:55 VanessaE +1
17:55 proller Megaf, read docs first
17:55 sfan5 could we stop doing non-constructive things?
17:55 proller i cant explain ~8 month of optimize process
17:55 Megaf That would be great
17:56 Megaf all you done so far is day freeminer is better and my hardware is slow
17:56 proller and first steps was ignored here
17:56 Megaf why dont you discuss on how you can improve minetest?
17:56 Megaf did anyone at least see my configuration??
17:56 Megaf do you have suggestions?
17:56 sfan5 <Megaf> did anyone at least see my configuration??
17:56 * sfan5 did
17:56 celeron55 Megaf: i checked a thing from it and noted that you already had the thing
17:57 Megaf celeron55: it was working fine, till I added builtin_items mod and a specific player joined
17:58 sapier do you really need that mod?
17:58 sapier who's author of that mod?
17:58 * Megaf will optimize a little more is config and disable builtin_items
17:58 Megaf sapier: its really useful, it cleans droped objects
17:58 Megaf something that minetest should do on its own
17:58 Megaf instead of just freezing on /clearobjects
17:58 sapier maybe it's doing something wrong, can you try rba's work?
17:58 Megaf again, minetest is slow by design
18:00 proller and designed for singleplayer without mods
18:00 sapier megaf we're working at it but we rome wasn't built on a single day too ;-)
18:00 Megaf from my own experience I can guarantee that it would be easier if you start from scratch instead of redoing parts of the code
18:01 Megaf First thing you need is a solid multi threaded base
18:01 Megaf with a very basic game
18:01 Megaf minetest 0.3 was pretty fast
18:01 Megaf what changed from there to here?
18:01 sapier but it's architecture was flawed in same way
18:01 sapier mods
18:02 sapier 0.3 didn't have any mod suppor
18:02 sapier t
18:03 sapier and that's exactly what you mentioned "works without mod" ... for my tests minetest 0.4 without mods is as fast as 0.3 ... maybe even a little bit faster
18:04 nore joined #minetest-dev
18:08 celeron55 designed for singleplayer without mods
18:08 Calinou <Megaf> minetest 0.3 was pretty fast
18:08 Calinou zero modding support
18:08 celeron55 ^ this is very true
18:08 Calinou you could play multiplayer too…
18:08 Calinou there was ban support and such, you could run decent-sized server
18:08 Calinou no rollback though, but rollback is currently disabled by default
18:09 celeron55 or actually, yes, multiplayer indeed
18:10 celeron55 0.3 still exists if someone wants to use it or fork it; it's not like it has disappeared or anything
18:10 grrk-bzzt joined #minetest-dev
18:10 * VanessaE mumbles something about minetest-classic
18:11 Calinou minetest-reactionary
18:12 sapier well lets focus on improving performance of minetest, we most likely wont get the performance of a simple game while having a complex one ... but maybe we can get close to it
18:12 * celeron55 avoids advocating specific forks
18:33 BlockMen is #1315 a bug? i thought damage_per_second is thought to damage player only when he is INSIDE the node
18:33 ShadowBot https://github.com/minetest/minetest/issues/1315 -- No damage_per_second while sneaking on top of the node
18:36 Megaf ok
18:36 Megaf one player (me) and max lag is 2
18:37 Megaf I thing I got a good conf now
18:37 Megaf max lag 8
18:37 Megaf bag to normal =/
18:44 vifino joined #minetest-dev
18:45 Megaf Does anyone knows what this means? emergequeue_limit_total
18:56 tomreyn joined #minetest-dev
19:38 Jordach joined #minetest-dev
19:47 diemartin joined #minetest-dev
20:35 proller joined #minetest-dev
20:51 cheapie joined #minetest-dev
21:06 BlockMen left #minetest-dev
21:41 pro joined #minetest-dev
21:42 kaeza joined #minetest-dev
22:06 RealBadAngel #1330
22:06 ShadowBot https://github.com/minetest/minetest/issues/1330 -- Allow dropping single items from stack while holding sneak key. by RealBadAngel
22:06 RealBadAngel another items related tweak
22:10 Megaf_ joined #minetest-dev
22:34 RealBadAngel any comments on above?
22:46 iqualfragile joined #minetest-dev
23:36 Skittle joined #minetest-dev

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