Minetest logo

IRC log for #minetest-dev, 2014-03-01

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

All times shown according to UTC.

Time Nick Message
00:09 Naked that diamond screenshot looks amazing
00:12 Naked looks so god damn amazing
00:30 paramat joined #minetest-dev
00:33 paramat simplified and lightweight hardware lighting would be interesting ... if that is even possible
00:40 sapier joined #minetest-dev
00:47 sapier left #minetest-dev
01:05 rsiska joined #minetest-dev
01:45 werwerwer joined #minetest-dev
01:58 paramat left #minetest-dev
05:42 OldCoder joined #minetest-dev
05:59 ImQ009 joined #minetest-dev
06:13 troller joined #minetest-dev
06:14 nore joined #minetest-dev
06:31 RealBadAngel btw, Terasology now has oculus rift support
06:32 RealBadAngel next month im gonna order OR for me too
06:33 RealBadAngel nore, where can i find your mc map converter?
06:34 nore https://github.com/Novatux/mcimport
06:34 nore (not yet working very good, and since the map.sqlite export, no more metadata)
06:35 nore (a bit slow too)
06:35 RealBadAngel have you used SN's tbn reader?
06:35 nore I've tried... but I'm not good at coding C++
06:35 nore anyway, I have some time tomorrow, I will try to do it
06:35 nore gtg
06:36 RealBadAngel oh cmon, whoever is good at lua, c is a piece of cake
06:36 RealBadAngel ;
06:37 RealBadAngel nore, how to use it?
06:39 nore just run python3 mcimport.py input/world/path output/world/path (must alreasy exist, it will only create map.sqlite)
06:46 RealBadAngel ok, will try to convert imperial cathedral
06:48 nore it will take some time...
07:54 PenguinDad joined #minetest-dev
08:07 darkrose joined #minetest-dev
09:49 troller joined #minetest-dev
10:19 PenguinDad joined #minetest-dev
10:29 rsiska joined #minetest-dev
10:38 proller__ joined #minetest-dev
10:50 Calinou joined #minetest-dev
11:03 troller joined #minetest-dev
11:05 OldCoder joined #minetest-dev
11:35 PilzAdam joined #minetest-dev
11:37 tomreyn joined #minetest-dev
11:42 elmux joined #minetest-dev
11:42 elmux Whats the state of the libmtmap map interface ?
11:45 SmugLeaf joined #minetest-dev
11:47 sfan5 elmux: 0 percent done
11:47 elmux okay just have been curious
11:51 Exio4 joined #minetest-dev
12:17 Jordach joined #minetest-dev
12:58 troller joined #minetest-dev
13:16 blaise joined #minetest-dev
13:41 EvergreenTree joined #minetest-dev
13:41 EvergreenTree joined #minetest-dev
13:52 domtron joined #minetest-dev
14:41 PenguinDad joined #minetest-dev
15:28 hmmmm joined #minetest-dev
15:53 daswort joined #minetest-dev
15:55 rubenwardy joined #minetest-dev
16:05 salamanderrake joined #minetest-dev
16:06 sapier joined #minetest-dev
16:09 smoke_fumus joined #minetest-dev
16:14 Calinou joined #minetest-dev
17:16 domtron joined #minetest-dev
17:35 EvergreenTree joined #minetest-dev
17:54 Selat joined #minetest-dev
17:57 EvergreenTree joined #minetest-dev
17:57 salamanderrake joined #minetest-dev
18:12 Naked joined #minetest-dev
18:17 Selat What do you think about using float instead of int for cost in pathfinder? This will allow using 8- and 16-adjacency.
18:18 sapier I don't know did you do a benchmark yet?
18:19 sapier I'm not convinced your code will always return best path but I don't have time to proof this right now ;-)
18:20 sapier And I believe your code is faster then the previous version too ... yet I've seen multiple times code looking faster beeing slower on doing actual test so please test your code
18:25 Selat I was wrong - it not always returns best path. But in case of equal cost it'll return best path.
18:26 Selat The problem is that it's impossible to find best path without processing all vertexes.
18:27 Selat In general case.
18:27 sapier I know about this, that's why best path was an option previous. there may be situations where you really want the best path
18:27 Selat Your code also doesn't return best path.
18:27 sapier it does
18:28 sapier at least within the specified parameters
18:28 Selat No. You process limited area (m_limits). Best path can be out of this area.
18:28 sapier of course, but I'm not sure about your algorithm providing best path within the specified limits
18:29 sapier e.g. if best path is first run in wrong direction ... the standard case where a* with manhattan distance fails
18:34 Selat But all costs are hardcoded. And with such hardcoded costs it always returns best path.
18:36 sapier as I said for what I understand your algorithm at first glance I have some doubt but I can't proove it. As this is a minor issue once you provide benchmarks really proving your code is faster we can merge it.
18:37 Selat The problem is that in case of minetest finding path and finding best path are two completely different tasks. I think it's better to split pathfinder fin two classes each for solving one certain task.
18:39 sapier that's what the different algorithms was meant to do ... why two different functions having same parameters and same return value?
18:41 Selat The difference not in algorithms, but in used data structures.
18:42 Selat Array for best path, and std::set for just path.
18:44 Selat *std::map
18:44 sapier as I said do a benchmark ... it'd not be first time a compiler/the os did already optimize code way better you tried to do it manually
18:49 Selat I just don't want to waste my time. And you haven't anwsered about float for cost.
18:49 Selat Compiler can't optimize not tail recursion. So in any way your code should be rewritten.
18:53 sapier well no need to discuss about theoretical benefits, find out how much becomes real ;-)
18:55 Selat How to create global variables in lua? I'm new to lua.
18:55 xyz all variables are global by default
18:56 xyz > Global variables do not need declarations. You simply assign a value to a global variable to create it.
18:56 Selat And it should be defined out of any function?
18:56 xyz no, it can be defined anywhere
18:57 xyz if you need a local variable, declare it using "local" keyword
18:57 proller__ joined #minetest-dev
18:57 Selat Thanks.
18:57 sapier but same as in c/c++ avoid globals where possible
18:58 Selat I need this just for benchmark.
18:59 Selat sapier: What about float for cost? It's important, because to my mind it'll be a little bit slower than int.
18:59 Selat so it's important for benchmark.
19:00 nore https://forum.minetest.net/viewtopic.php?id=8489 <-- what about adding this as a new hud item?
19:00 sapier I don't know if float will cause a significant slowdown I suggest doing a benchmark now against old code and another one float <-> your current version
19:00 Selat ok
19:01 xyz nore: this can't be merged
19:01 sapier imho you should benchmark at least following usecases: simple pathfinding (multiple times), complex ones, simple one without existing path and a conplex example without a existing path
19:01 nore xyz: I didn't look at it, but what are the problems?
19:02 xyz nore: source code license
19:02 nore ?
19:04 xyz what's unclear here? it can't be merged because of his code license
19:04 nore ah, I saw... it is GPL
19:04 VanessaE so?  just ask him.
19:05 xyz yeah, this'll do
19:06 VanessaE I'm serious.  Ask the guy for permission to merge his code into Minetest core under the usual license therein.  either he agrees to dual-license it, or he doesn't.
19:07 xyz are you assuming I was joking or what was the reason that made you repeat yourself again?
19:07 VanessaE I assumed you didn't understand my point (wouldn't be the first time :P )
20:02 zat joined #minetest-dev
20:15 domtron joined #minetest-dev
20:16 sapier left #minetest-dev
20:18 Ritchie joined #minetest-dev
20:20 salamanderrake joined #minetest-dev
20:27 salamanderrake joined #minetest-dev
20:41 domtron_ joined #minetest-dev
21:01 Jordach_ joined #minetest-dev
21:10 proller__ 01:08:26: ERROR[ServerThread]: Server::ProcessData(): Cancelling: No player for peer_id=2043
21:10 proller__ 01:08:26: ERROR[ServerThread]: Server::ProcessData(): Cancelling: No player for peer_id=2037
21:10 proller__ 01:08:26: ERROR[ServerThread]: Server::ProcessData(): Cancelling: No player for peer_id=2040
21:10 proller__ 3 вшысщттусеы шт щту ешьу
21:10 proller__ 3 disconnects in one time
21:37 sfan5 proller__: just pasting stuff in here and expecting a fix isn't good
21:37 sfan5 your're a dev yourself, try to find the problem and try to fix it
21:38 proller__ revert sapier changes into connection.cpp ?
21:38 proller__ now i cant understand it
21:39 VanessaE I have never seen errors like this on my servers, proller.
21:39 VanessaE and I have hundreds of hours logged.
21:39 VanessaE are you using patch #1143?
21:39 ShadowBot https://github.com/minetest/minetest/issues/1143 -- Protocol version <= 22 compatibility mode + clean new client init by sapier
21:40 proller__ no, not changed connection from master
21:40 VanessaE actually more like thousands of hours.
21:40 VanessaE that's why.  you need that patch or you're gonna see errors.
21:41 xyz http://irc.minetest.ru/minetest-dev/2014-02-27#i_3600196
21:41 xyz wtf
21:41 xyz why isn't it merged if you "need" it
21:42 VanessaE xyz:  because it hasn't been established if the way forward is to keep compatibility with the old protocol or to totally trash it all in favor of something new that won't ever work with 0.4.9-and-prior
21:42 VanessaE until that question is answered, 1143 remains in limbo
21:43 xyz but you said "you need that patch or you're gonna see errors."
21:43 xyz so the game fucking doesn't work
21:43 proller__ it work, but only 3-5 minutes
21:43 proller__ then reconnect
21:44 xyz of course expecting every revision to work as good as stable releases is wrong, too
21:44 VanessaE xyz: apparently most everyone else isn't actually having problems
21:44 xyz but when there's a patch that fixes everything
21:45 Jordach_ i find it funny how i'm using the stable and it hasn't had problems
21:45 VanessaE and afaik there is no significant problem using stable, either
21:45 sapier joined #minetest-dev
21:45 VanessaE hah, speak of the devil!
21:45 VanessaE it's all your fault, sapier :)
21:46 sapier what did I do this time? ;-)
21:46 VanessaE sapier: you failed to merge 1143 :P
21:46 sapier #1143
21:46 ShadowBot https://github.com/minetest/minetest/issues/1143 -- Protocol version <= 22 compatibility mode + clean new client init by sapier
21:46 VanessaE sapier: re: proller's usual bitching about random network bugs :)
21:47 sapier well :-) 1143 will break compatibility completely so I wasn't that eager prior having a working android client
21:47 sapier ... thought it'd be a week only to get those ones around to a solid state ... I was wrong
21:47 VanessaE break it with what?  remember, it's working fine on my servers, other than these old-ass tablet clients that aren't properly managing their caches
21:48 sapier did anyone find the reason why minetest is limited to 1 fps in multiplayer? :-)
21:48 sapier no no 1143 will be merged in a break all way I will not add the compat code to master
21:48 VanessaE eep
21:48 VanessaE well there's your answer, xyz
21:49 sapier basicaly it's gonna be 0.5.0
21:49 PenguinDad joined #minetest-dev
21:49 sapier that's what poll result requests
21:49 xyz VanessaE: yet you told this fixes this error
21:49 sapier I'd prefere adding compat code but if I remember correct the poll was very very clear about this
21:49 sapier 1143 contains compat code in it's current state
21:50 xyz if you don't need compatibility then it's a good time to switch to enet
21:50 sapier if you have a reliable enet version ?
21:51 xyz what do you mean?
21:51 sapier oh well and benchmarkable of course because the one I have was half as fast as tcp variant
21:51 xyz so are you going to switch to tcp completely?
21:51 sapier but enet results have been strange so no idea if they are correct
21:51 xyz that's just insane
21:51 sapier actually minetest uses udp in a tcp like way
21:52 sapier the difference isn't as big as it seems to be ... but I don't wanna start this discussion now
21:52 sapier xyz how did you solve the performance issue on android multiplayer?
21:53 xyz what performance issue?
21:53 VanessaE xyz:  said, and I quote:  "that's why.  you need that patch or you're gonna see errors." ...  I didn't say it "fixes" anything.  I've got thousands of server hours and lots of happy users on multiple platforms and versions.  The only glitch any of them can report is strictly a problem with their clients not showing all logged-in users, and we can't fix that server-side.
21:53 VanessaE I said, and*
21:53 sapier well I've got 10 fps in singleplayer and 1 in multiplayer :-)
21:53 xyz alright alright
21:54 xyz sapier: what's your device?
21:54 sapier a cheap android tablet
21:54 xyz throw it away
21:54 sapier well that's what you allways say instead of looking for the bug
21:55 xyz sure
21:55 sapier having 10 times as much fps in singleplayer already prooves it's not the pure cpu power
21:55 xyz do tell if you find it
21:56 xyz it surely will be a fun laugh
21:57 sapier btw why didn't you cleanup android implementation? Usually you do rewrite everything but things as ugly as celerons sailfish prototype don't bother?
21:57 xyz uh? sorry, can you rephrase this?
21:59 xyz how's celeron's sailfish patch related? I didn't even touch it
21:59 sapier exactly ... his patch was crap
21:59 sapier especially making touchscreengui a derived class of MyEventReceiver
22:00 sapier while having MyEventReceiver still in code doing same thing
22:01 sapier basicaly RealInput handler sometimes accessed MyEventReceiver sometimes touchscreengui, both holding (inconsistent) copys of gui button state
22:01 xyz I don't really understand, it's derived from InputHandler, where's MyEventReceiver there?
22:02 xyz and most of hacks you're complaining about are probably mine anyway
22:02 xyz feel free to rewrite and MAKE IT BETTER
22:02 xyz I kinda put it on hold now, doing other things
22:02 sapier no it's not your fault, it's in sailfish as well as in stu's build so this is in the starting base
22:03 xyz and gues who wrote the starting base
22:03 sapier thought it's been celerons sailfish?
22:03 xyz yeah sure
22:04 sapier but It's tough stuff, your irrlicht fixes have been quite usefull
22:04 xyz anyway
22:04 sapier took me some time to realize all the black magic done in various locations
22:04 xyz lrn2git blame
22:04 xyz what magic?
22:05 sapier usually I consider looking who broke it to be wasted time
22:05 xyz what exactly is broken?
22:06 xyz yet you consider blaming wrong people an acceptable practice?
22:06 sapier well I fixed most of it ... one thing I don't know about is wchar <-> mbs conversion ... androids fcts seem to do nothing at all
22:06 sapier and minetest doesn't have a portab layer to switch those functions at a single location ... guess that's next to add
22:07 xyz where can I take a look at it then?
22:07 sapier I've got some debug code in yet so I don't wanna push it now but I intend to clean it tomorrow to make it available to public
22:09 xyz why not push it?
22:09 xyz I'd really like to check it out
22:09 sapier lemme have a look how much crap is in there ;-)
22:10 xyz who cares
22:10 sapier I do as I have to clean it to be able to merge
22:10 sapier commits that ill shaped are not mergeable to master
22:10 sapier -ill+miss
22:10 xyz I still don't see any problem in pushing your changes
22:11 xyz it's your personal repo
22:11 xyz I mean, you're saying like "I fixed this and that but you're not allowed to see it"
22:11 xyz what's the point in telling us you fixed this stuff then?
22:12 sapier as I said it's more easy for me to clean it at once then looking at 100 commits finding the one I need
22:12 xyz so what exactly did you fix then?
22:12 xyz if you aren't willing to show the code, let's talk about it
22:12 sapier but if you're so eager to see the current (beta) state I'm gonna do some cleanup and show push it
22:13 xyz so now I get neither
22:13 xyz good
22:14 sapier but it's unfinished lots of things are started but not yet complete
22:15 xyz so what exactly did you change? can you answer that?
22:15 sapier major change is touchscreengui is a translation entity only not replacing the original code but beeing used to mimicry old input
22:16 sapier a initial version of a simplified mainmenu is added too
22:16 xyz what is simplified?
22:17 sapier 3 tabs, no multiworldsupport no texturepackselection, smaller to fit on screen, no game selection
22:17 sapier basicaly stripped everything but the basic things
22:17 xyz but what's the point?
22:17 sapier it's simple
22:18 xyz so what is it?
22:18 sapier I don't understand the question?
22:18 xyz well you said the point is simple or what?
22:18 xyz or you mean, it's more simple this way?
22:19 sapier it IS more simple but of course it's limited too
22:19 xyz and since all android players are complete idiots (which is true actually) they won't be able to comprehend the genius of our main menu
22:19 xyz I don't think this is very nice
22:19 xyz anyway
22:19 xyz if you're willing to work on stuff, how about making this shit work nicely with high dpi displays?
22:20 sapier I more thought about cutting the menu to the main usecases on a android device ... but as you know for sure there's a setting to select which menu to show ;-)
22:20 xyz I have a feeling that's more important than another useless refactoring
22:20 sapier so you can use the original mainmenu on android too (at least once the checkbox is added)
22:20 xyz since this stuff is still unfinished there's no need to refactor anything
22:21 xyz and GUI really looks awful on high dpi displays right now
22:21 sapier you need to refactor everything as you have to rewrite basic parts of gui ... I assume most things would be fixed if guiformspecmenu would honor dpi
22:22 sapier but first I want to finish the basic features
22:22 xyz what features?
22:22 sapier user interaction as well as reasonable performance
22:23 sapier right now I can't use all those shiny old style menus left in minetest
22:23 sapier haven't had a look where to fix this
22:23 xyz what menus?
22:23 sapier e.g. pause menu
22:25 xyz good luck either way
22:26 xyz btw, did you read android users' comments to buildcraft on google play page?
22:26 sapier thanks ... that android platform is creapy
22:26 xyz nah, it's actually kinda nice
22:26 sapier not yet, is there something interesting?
22:26 xyz it's removed from google play now so all's gone
22:26 xyz it's just
22:27 xyz after I read it I got into a state of disappointment
22:27 sapier what was the main issue?
22:27 sapier or the main issues
22:27 xyz oh it's hard to explain
22:27 xyz I just felt that those users don't deserve this
22:29 sapier :-) you need to learn to love that feeling if you wanna continue oss ... the most important work is the work you don't get any reward because no one realizes it's been done
22:30 xyz no, I tried it with some other stuff and I did get reward
22:31 xyz it's just a weird combination of game genre, etc, which attracts this kind of audience
22:31 sapier there's always some exception from the rule
22:35 xyz you're only saying this because you haven't seen the reviews
22:37 AllegedlyDead joined #minetest-dev
22:37 sapier maybe but I can imagine how bad they have been
22:37 xyz no no no, they've been, like, at least 10 times worse than you just imagined
22:38 sapier I can imagine very very bad comments ... but I can't even blame them, assuming buildcraft was twice as good as what I made by now it'd been far from end user proof
22:41 Sokomine sapier: move away from spawn on vanessas vanilla server. for reasons unkown, i lately had trouble at spawn on the vanilla server with stus build. further from spawn - no problems. it is very odd. something like that happend several versions ago with the desktop version of minetest on some servers as well. might still be another issue...
22:45 xyz ok I think I had some pics somewhere but they're in russian so you'll have to learn this language in order to understand how awful they really are
22:46 sapier hmm sokomine you're right
22:46 sapier guess that's another variant of vanessaE's well known "engine is broken" issue
22:47 Sokomine xyz: the majority of players attracted to servers with the android port does not do much useful things on a server. i'm afraid that will always be so. there'll be some players amongst the mass that will be worth it. and although hardly anybody will ever say thank you directly, some will show indireclty
22:48 sapier I guess the best thank you may ever get is a high download rank
22:49 Sokomine xyz: it will be difficult to understand the comments then
22:49 Sokomine sapier: just run away from spawn :-) a few meters may be enough. say, 10-20 m from spawn - and you may live. with decent fps
22:50 Sokomine when i tested stu's first bulid, that issue was not there. it happend later. no idea what it is
22:51 sapier could be the mesh reloading
22:51 Sokomine mesh reloading?
22:52 sapier I have a gprof trace from spawn maybe I can see what's going on if I compare it to another one not at spawn
22:52 Sokomine that's a good idea
22:52 sapier I didn't manage to get address sanitizer work on android platform at all but profiling did work at once ... that platform is strange
22:52 Sokomine something strange has to happen there. as the vanilla server does not have anything suspicious (it's mostly minetest_game), i don't see what might cause issues
23:49 sapier joined #minetest-dev

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