Minetest logo

IRC log for #minetest-dev, 2014-08-12

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

All times shown according to UTC.

Time Nick Message
01:13 Zeno` joined #minetest-dev
01:30 nore_ fu-fu: it is the 16x16x16 mapblock that contains pos that is forceloaded
01:30 kahrl joined #minetest-dev
02:49 zat joined #minetest-dev
02:51 Miner_48er joined #minetest-dev
03:29 fu-fu nore_: Thankyou
03:44 MaxCohen joined #minetest-dev
03:45 Zeno` joined #minetest-dev
04:24 RealBadAngel hi
04:25 Zeno` hi
04:27 RealBadAngel hi Zeno`
04:29 RealBadAngel i can see you had problems with squashing those commits
04:30 RealBadAngel hmmmm, around?
04:35 Zeno` oh yeah I did
04:36 Zeno` I squashed into the wrong commit heh
04:36 alexxs joined #minetest-dev
04:36 Zeno` the commit I picked to sqash and the commit I picked to "pick" should have been the other way around
04:36 Zeno` oops
04:53 Zeno` joined #minetest-dev
05:00 RealBadAngel maybe best way could be just to apply changes manually on the clean file?
05:01 Zeno` maybe
05:02 Zeno` and do a whole new commit?
05:03 Zeno` All the commits actually exist in this one: https://github.com/Zeno-/minetest/commit/8f57e0e1d2b9b9fb6723083d894cab6518a0b44f
05:04 Zeno` So they're not lost... just not where I really wanted them to be
05:08 Miner_48er joined #minetest-dev
05:11 RealBadAngel even when you remove the tree commit will be still here
05:14 Zeno` I have no problem starting a new branch and committing from scratch; the real problem is I addressed two open issues in one commit (because the second commit was adding 1 line to l_env.cpp to set a default value for random_level)
05:15 Zeno` I hindsight I would have done it differently heh
05:43 Hunterz joined #minetest-dev
06:48 werwerwer joined #minetest-dev
07:15 init joined #minetest-dev
07:18 Amaz joined #minetest-dev
08:01 RealBadAngel kahrl, here?
08:36 deltib joined #minetest-dev
08:43 casimir joined #minetest-dev
09:06 Calinou joined #minetest-dev
09:16 Eater4 joined #minetest-dev
09:35 Amaz joined #minetest-dev
09:39 proller joined #minetest-dev
09:52 kaeza joined #minetest-dev
10:16 Amaz joined #minetest-dev
10:36 Zeno` quiet
10:58 PenguinDad joined #minetest-dev
11:13 proller joined #minetest-dev
11:16 ImQ009 joined #minetest-dev
11:35 proller joined #minetest-dev
11:36 Exio joined #minetest-dev
11:49 Jordach joined #minetest-dev
11:56 fu-fu joined #minetest-dev
13:12 VanessaE joined #minetest-dev
13:22 NakedFury joined #minetest-dev
13:28 Amaz joined #minetest-dev
13:35 proller joined #minetest-dev
14:06 rubenwardy joined #minetest-dev
14:14 rubenwardy_ joined #minetest-dev
15:22 proller joined #minetest-dev
15:28 Calinou joined #minetest-dev
15:34 eugd joined #minetest-dev
15:40 eugd why not have MAP_GENERATION_LIMIT as a server variable?
15:41 eugd i've seen mention of mods to limit generation, shouldn't that be a basic engine feature?
15:42 eugd If they're not working through the engine I can only think they're probably very inefficient (just forcing all unmade blocks to be air?)
16:03 zat joined #minetest-dev
16:06 rubenwardy joined #minetest-dev
16:06 Hunterz joined #minetest-dev
16:07 Anchakor_ joined #minetest-dev
16:07 eugd MAP_GENERATION_LIMIT as a server setting?
16:12 VanessaE why?
16:12 VanessaE the map is generated only as needed
16:13 VanessaE even with mods that generate map stuff.
16:14 VanessaE a player explores a new area, the engine handles generating only the blocks within range of the player (or tells the mods to do so).
16:14 VanessaE it doesn't just generate "everything"
16:14 Jordach VanessaE, he means upto a point, the mapgen will just generate nothing, like the walls of void, but closer
16:15 Jordach (eg making the 31km invisi walls at <x>KM instead)
16:16 PenguinDad A mod could already do this without a big performance impact
16:16 eugd VanessaE, this is currently in the engine as a constant
16:16 eugd set to 31000
16:16 VanessaE eugd: that's just as far as the engine will go
16:16 VanessaE that doesn't mean it will go there always
16:17 VanessaE it just means it won't go *past* there.
16:17 eugd i was thinking it would be better to do it in engine than with mods
16:17 eugd yes i understand
16:17 eugd i've heard mention of mods to shrink this further
16:17 eugd which makes sense
16:23 celeron55 if you do it in the engine, then people are going to request eg. other shapes and whatever special stuff which won't be added to the engine
16:24 celeron55 so if it's possible to do in mods without issues, then it should be because that is the extendable way
16:24 celeron55 (this applies to everything)
16:27 eugd haha i was thinking of it as a likely necesarry first step to other 'special stuff'
16:32 eugd I'm curious if there is any technical barrier to it at all
16:32 eugd just adding it to the server config file and changing all references to that
16:49 grrk-bzzt joined #minetest-dev
17:00 zat joined #minetest-dev
17:03 ^v joined #minetest-dev
17:24 celeron55 probably not
17:30 Calinou the idea behind making it a variable is that you can use it to not have too large maps (file size)
17:30 Calinou and make it slighly larger (32000?)
17:32 Miner_48er joined #minetest-dev
17:34 Krock joined #minetest-dev
17:40 Taoki joined #minetest-dev
17:56 EvergreenTree joined #minetest-dev
18:02 VanessaE joined #minetest-dev
18:04 VanessaE_ joined #minetest-dev
18:18 cg72 joined #minetest-dev
18:23 khonkhortisan joined #minetest-dev
18:31 pitriss joined #minetest-dev
18:31 casimir joined #minetest-dev
18:37 proller joined #minetest-dev
18:52 eugd what is the "chunksize" field in map_meta.txt?
19:03 Krock 5 = 80*80*80 nodes get generated at once
19:03 Krock 5 * 16 = 80
19:03 Krock set it to 4 for faster map generation and copputions of already existing maps
19:08 luizrpgluiz joined #minetest-dev
19:08 luizrpgluiz left #minetest-dev
19:22 eugd "copputions"?
19:23 PenguinDad eugd: I guess he meant corruptions
19:24 sapier joined #minetest-dev
19:24 sapier VanessaE try disabling farming
19:25 sapier I see abm spikes up to almost a second on initial join
19:27 VanessaE_ ?
19:27 VanessaE_ in *farming*  ?!
19:27 sapier remember the profiler we did talk about?
19:27 VanessaE_ yes of course
19:27 sapier I've got a prototype only tracing some calls, including abms
19:29 sapier https://gist.github.com/sapier/0407cc49376b0ef68ed6 see the maximum value for farming abm ... it's almost a second ... and that's a new world
19:29 sapier those are "per step" values means one server step is delayed by that much time
19:30 VanessaE_ ouch
19:30 sapier as you can see the average value for farming is quite good seems like a login spike only
19:37 VanessaE_ any clue just what's going on there?
19:37 sapier oops
19:37 VanessaE_ oops?
19:37 sapier those spikes aren't on join only
19:37 sapier they happen every now and then
19:38 sapier I'd guess for some reason abm's aren't spread good enough
19:39 VanessaE_ are they stacking due to mapgen lag maybe?
19:41 sapier possible but as you see in that paste on_generated isn't laging ... but I don't trace all possible things right now
19:42 VanessaE_ true
19:46 sapier grrr ... ShadowNinja could you please fix the error in errorhandling caused by a bug in some entity callback? ;-)
20:00 XDdrummer joined #minetest-dev
20:01 XDdrummer Any developers in here at the moment?
20:03 sfan5 yes
20:04 XDdrummer Just wondering if there's any sort of process to joining the dev team.
20:05 sapier onle the usual process how to contribute to oss projects
20:06 sfan5 start contributing
20:06 sfan5 e.g. writing code, docs or other stuff
20:06 sfan5 and if you've contributed enough
20:06 XDdrummer Mkay, thank you :p
20:07 sapier ahh well one special thing ... on contributing to minetest you agree to be flamed every now and then by everyone
20:08 XDdrummer xD
20:10 VanessaE_ just one other thing:
20:10 sapier https://gist.github.com/sapier/b2885fccf7e77800451a
20:11 VanessaE_ make an account on github and don't give us any shit about hating their terms or whatever ;)
20:11 * VanessaE_ pokes at Zefram ;)
20:11 sapier compared to farming even mobfs spawning is lightweight ;-P
20:11 VanessaE_ sapier: are you sure the figures are correct?
20:12 sapier they're not complete, thus percentage values may be slightly of but abolute us values are correct
20:13 sapier wow ... those chicken are quite hungry :-)
20:13 VanessaE_ heh
20:13 VanessaE_ dare I ask you to run Dreambuilder through it? ;)
20:13 sapier yet it's 0.7ms delay ;-)
20:13 VanessaE_ (grab build #20140812-1612)
20:14 sapier can you give me a link=
20:14 VanessaE_ https://github.com/VanessaE/dreambuilder_game
20:14 sapier btw changes to minetest are 99.9% pure lua
20:14 sapier the only thing I had to do in c++ was providing a high precision time source
20:14 VanessaE_ ice
20:14 VanessaE_ nice
20:15 sapier I could've done using luasocket too but I'd like this to be usefull without additional software
20:15 sapier is dreambuilder complete or do I need default game too?
20:15 VanessaE_ it's complete.
20:15 sapier ok
20:16 sapier btw last gist was simplified display, I can show details what in mod is using most time too
20:16 sapier but the output was already way to long for mobf :-)
20:18 sapier what is this game doing?
20:19 sapier https://gist.github.com/sapier/268212b0fe1a2ad277ef
20:19 nore_ joined #minetest-dev
20:19 sapier git master
20:19 sapier 12s stall in globalstep?
20:19 VanessaE_ mapgen probably
20:20 nore_ joined #minetest-dev
20:20 sapier no on_generated by now ... but I'm stalled right now too :-) so could be next
20:20 VanessaE_ wait, you're still in minetest_game?
20:20 sapier no dreambuilder
20:21 VanessaE_ I have no idea :P
20:22 sapier that list should be printed to console every 10s but hasn't been printed for about 5 min now ;-)
20:22 VanessaE_ it shouldn't be doing anything at the start but a shitload of mapgen stuff
20:22 sapier yes I did start a new world
20:22 ImQ009 joined #minetest-dev
20:23 VanessaE_ hm, maybe the ambiance mod?
20:23 sapier I try without my tracing addons to be sure
20:24 VanessaE_ either that or snowdrift, those are the only two that should be able to do anything at the start of a new world.
20:24 VanessaE_ (if it's ambiance, I'll remove it.  it was just added today and I'm unsure if I want to keep it)
20:25 sapier ok better without tracing
20:25 VanessaE_ ?
20:25 sapier doesn't stall completely
20:26 VanessaE_ wow.
20:26 VanessaE_ it does raise the server's CPU needs a little bit. but only briefly.
20:27 VanessaE_ 22824 minetest  20   0 1892m 380m  49m S  24.8  4.8  31:45.52 /usr/bin/minetestserver --worldname Survival_World --config /home/min
20:27 VanessaE_ (there are three of us on that server right now)
20:27 VanessaE_ ok, it's not ambiance then.   must be something else
20:28 sapier https://gist.github.com/sapier/268212b0fe1a2ad277ef on world generated without
20:28 sapier worst step is delayed by 16s
20:28 VanessaE_ well all those on_generated calls is unsurprising.
20:29 sapier ok forget about it now worst one is 28 seconds
20:29 VanessaE_ in?
20:31 VanessaE_ I really wonder what could be causing huge stalls in globalstep like that
20:31 VanessaE_ could you please delete ambiance and try again?
20:32 sapier according to this traces ambience mod is almost irrelevant
20:32 VanessaE_ *looks at grep output*
20:32 VanessaE_ snowdrift?
20:32 sapier moretrees and ferns  are biggest single laggers ... sorry VanessaE
20:33 VanessaE_ er, moretrees doesn't use globalstep :P
20:34 sapier those are on_generated or abm lags
20:34 VanessaE_ oh ok
20:34 VanessaE_ so just ignore those
20:34 VanessaE_ at least the ones attributed to builtin
20:35 sapier but there's at least a bug in displaying values too as I don't believe multiple entries to have exactly same value
20:35 VanessaE_ are you running this with luajit?
20:35 sapier I don't know
20:36 sapier let me check
20:37 sapier no I don't use luajit
20:37 VanessaE_ try with please
20:37 sapier ok ... hopfully tracing still works ... hmm it's lua only so it should work
20:37 VanessaE_ should work I'd think
20:38 sapier still you should improve moretrees and ferns ;-)
20:38 VanessaE_ I plan to.
20:38 VanessaE_ I have to rewrite plants_lib in ways that will affect everything that depends on it, so today I've imported ferns, dryplants and undergrowth into plantlife
20:39 sapier 10s are way to much for a single step ;-) ... wonder why ppl only complain about mobf which doesn't cause lag near to those values ;-)
20:39 VanessaE_ which *should* cover all of the otherwise-unsupported mods
20:39 VanessaE_ however:
20:39 VanessaE_ I need YOU to help:
20:39 VanessaE_ moretrees needs the engine's help specifically:  the spawn_tree() call is dog slow.
20:40 VanessaE_ when last I was told, because the lighting code therein is slow.
20:40 VanessaE_ this needs fixed before moretrees can improve too much
20:40 VanessaE_ but I will do what I can from my end.
20:41 sapier not sure if wouldn't be more easy to do the spawning different but hard to tell without knowing details
20:41 VanessaE_ another thing that would greatly benefit plantlife/plants_lib is a search-for-surfaces algorithm in the engine.  THAT is the biggest bottleneck.
20:42 VanessaE_ if I could just call minetest.find_surfaces_in_area(minp, maxp) --> returns table of nodes that have air above
20:42 sapier hmm that one is sitting around for ... don't know ... about a year?
20:42 VanessaE_ eh?
20:44 VanessaE_ (surfaces, preferably with a filter option e.g. only stone with air above, etc)
20:44 sapier https://github.com/minetest/minetest/pull/640
20:44 VanessaE_ ohyeah
20:45 sapier almost 1.5 years
20:46 sapier https://gist.github.com/sapier/86ad62c58531b865a4a8  luajit values
20:46 sapier worst case is now 23 seconds
20:47 VanessaE_ well 5 seconds faster
20:48 sapier biggest single laggers are (in order of delay) ferns moretrees builtin flowers_plus trunks dryplants woodsoils
20:48 sapier everything else is below 1s
20:48 VanessaE_ what's in builtin that's lagging though I wonder?
20:49 sapier I don't know seems something on init it's on globalstep and done on init only ... maybe it's loading of mods
20:49 VanessaE_ (mind you this is older code, so compare it to pre-minetest_next.  I don't track current minetest_game)
20:49 VanessaE_ oh ok
20:50 VanessaE_ who cares about init/startup then
20:50 sapier actually that can't be
20:50 sapier no ... it can be
20:50 sapier ... hmm guess I don't know
20:50 VanessaE_ maybe you better track register_node/register_craft* calls ;)
20:51 sapier those should be accounted to the mods
20:51 sapier builtin can only be initializing builtin data
20:51 sapier don't know why this is that much
20:55 proller joined #minetest-dev
21:02 VanessaE_ no clue
21:10 sapier https://github.com/sapier/minetest/tree/luamod_tracing in case you wanna try yourself
21:10 sapier it's not complete yet but most common things are already traced
21:11 VanessaE_ I'm afraid to try :P
21:15 VanessaE_ all that time spent in builtin....could it be the tracing code counting itself?
21:16 sapier console printing maybe
21:16 sapier but not those 12s in dreambuilder
21:16 VanessaE_ naw
21:16 VanessaE_ I just meant that one particular part.
21:17 VanessaE_ I'm pretty sure I know what needs done to speed up the stuff that depends on plants_lib
21:17 sapier tracing doesn't count itself thus it shouldn't be included
21:23 sapier I wonder why I haven't written this tool before it's quite small and provides a lot of information
21:23 VanessaE_ you didn't have me here egging you on? ;)
21:24 nore_ joined #minetest-dev
21:24 sapier possible ... but I guess I got annoyed about mobf beeing blamed for other mods delays
21:24 VanessaE_ haha
21:25 sapier about 30 active mobs and mobf isn't using more then 6ms
21:25 sapier 3 of them used by about 15 chicken :-)
21:25 VanessaE_ heh
21:26 * VanessaE_ looks at plants_lib's code and considers the rewrite in question...and shudders,
21:26 sapier trapped within a very small area thus doing permanent collision avoidance
21:27 sapier you could have a look at adv_spawning maybe it's less work to improve it to fit your requirements
21:30 VanessaE_ I'll figure it out, no worries.
22:21 kaeza joined #minetest-dev
22:22 fu-fu joined #minetest-dev
23:23 grrk-bzzt joined #minetest-dev
23:23 grrk-bzzt joined #minetest-dev
23:50 Exio joined #minetest-dev

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