Minetest logo

IRC log for #minetest-dev, 2015-05-15

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

All times shown according to UTC.

Time Nick Message
00:59 prozacgod joined #minetest-dev
01:08 Zeno` joined #minetest-dev
01:22 rickmcfarley joined #minetest-dev
01:29 Zeno` est31, is SRP login working for you on Android?
01:29 est31 umm
01:29 est31 lemme check
01:29 est31 is it not for you?
01:30 Zeno` No, but I don't know if it's because the lib is not being installed properly or a different issue
01:30 Zeno` the lib certainly downloads and builds
01:31 est31 do you know that srp is currently not activated?
01:31 Zeno` I revert your commit and things work again
01:32 Zeno` yeah, I think it's just not finding the lib and... hmm, it's weird
01:32 est31 but why does it build then
01:32 Zeno` building properly != working :P
01:33 est31 yes but usually building == libs found
01:33 est31 or building => libs found
01:33 est31 (other implication doesnt hold :p)
01:34 Zeno` well it's not a static lib, so no...
01:34 Zeno` anyway, this is what I get: http://dpaste.com/2HTVZW1
01:35 est31 vorbis and ogg??
01:35 Zeno` what about them?
01:36 est31 nvm just a warning
01:36 est31 not the cause of the error
01:36 Zeno` oh, yeah... I guess some libs can have some kind of init callback
01:39 est31 ok I get it too
01:39 OldCoder joined #minetest-dev
01:40 Zeno` Where are libs supposed to be "installed"?
01:40 Zeno` ShadowNinja knows
01:41 Zeno` I can never remember heh
01:41 Zeno` although I'm not sure that's even this issue
01:43 est31 libgmp.so  is inside the apk
01:44 est31 the funny thing is that it complains about minetest.so, and doesnt even mention libgmp.so
01:45 Zeno` yeah I know :/ But it's definitely your commit (try it if you don't trust me ;))
01:45 est31 nono
01:45 est31 trust you
01:45 Zeno` Stupid android
01:45 est31 yes :p
01:46 est31 is it the srp commit or the gmp commit?
01:48 est31 the gmp one
01:51 est31 my guess is that I forgot to add the library somewhere
01:52 est31 yes I did
01:53 chchjesus joined #minetest-dev
01:53 Zeno` well I added it and it didn't *seem* to make a difference but I may have done it wrong so I'll wait for your attempt :)
01:56 est31 yea ...
01:56 est31 perhaps I've done it wrong too :)
01:56 est31 wrongly *
02:12 Zeno` we both sound, umm, very confident!
02:12 Zeno` I dunno... I just added gmp where all the other libs are in MtNati...java
02:13 est31 yes like me
02:13 est31 my second try was to put it first
02:13 est31 because this guy writes https://github.com/Rupan/gmp
02:14 est31 "The GMP library must be loaded prior to other dependent modules.  This is an apparent shortcoming / limitation of Android."
02:14 est31 didnt work either
02:15 est31 so yes, it should be there, but it shouldn't crash.
02:19 est31 ah I have something
02:19 est31 nice nice
02:20 est31 I've added System.loadLibrary("minetest"); just for the lulz
02:20 est31 then it actually produced an error message: dlopen("/data/app-lib/net.minet​est.minetest-1/libminetest.so") failed: dlopen failed: could not load library "libgmp.so.10" needed by "libminetest.so"; caused by library "libgmp.so.10" not found
02:20 est31 nice nice
02:22 Zeno` confirmed at least :D
02:22 Zeno` where should libgmp.so.10 be located on the device?
02:22 est31 it isn't
02:23 Zeno` why?
02:23 est31 I guess its some "resolve the symlink and then take the filename error"
02:23 Adimgar left #minetest-dev
02:23 Zeno` why isn't it on the device?
02:23 est31 because libgmp.so is.
02:23 Zeno` oh, you know what I mean :P
02:24 Zeno` where is libgmp.so located :P
02:24 est31 in the apk
02:24 est31 lib/armeabi-v7a/somewhere
02:24 Zeno` but it should be on the device somewhere after the APK is unpacked... shouldn't it?
02:24 est31 yes
02:24 est31 I mean its successfully loading libgmp.so
02:25 est31 just somehow libminetest thinks it wants libgmp.so.10
02:25 Zeno` oh duh
02:25 Zeno` didn't read the whole error
02:25 est31 we have to teach libminetest to want libgmp.so
02:26 Zeno` teach it/.... lol
02:26 Zeno` :D
02:26 est31 :p
02:38 est31 btw this shows the problem very nicely: arm-linux-androideabi-readelf -d libs/armeabi-v7a/libminetest.so
02:40 Zeno` I didn't even know you could do that heh
02:51 est31 ok great found a fix
02:52 Zeno` est31, https://github.com/minetest/minete​st/blob/master/src/game.cpp#L3003
02:52 est31 thanks to that dude above
02:52 est31 Zeno`, ?
02:52 Zeno` I am removing the exclude bits and the if (so only lines 3008 and 3009 will remain).
02:53 Zeno` I don't recall my reasoning behind the exclude bits in the first place. I think I mistakenly thought that without them things like digging etc would not work but this is not the case
02:53 Zeno` I do remember thinking that people would fall if cliffs if they were sneaking, but this is not the case either
02:53 Zeno` I.e. I should never have added them
02:54 est31 ok
02:54 Zeno` they do nothing except make people annoyed lol
02:54 Zeno` e.g. Wayward_One
02:54 Zeno` what was the Android problem?
02:54 est31 https://github.com/Rupan/gmp/blo​b/master/compile-gmp-arm.sh#L30
02:54 Zeno` whoa
02:55 est31 added that recompiled libgmp and it works
02:55 Zeno` nice :)
02:55 est31 totally straightforward
02:55 Zeno` once you know :)
02:55 est31 ok its understandable once you know
02:55 Zeno` I like how GMP use -pedantic in their build flags
02:56 Zeno` don't ever use -pedantic on minetest
02:56 est31 lol
02:56 Zeno` something like 10000 warnings :/
03:01 Hijiri joined #minetest-dev
03:13 est31 pushing the fix
03:13 Zeno` ok
03:14 Zeno` I will push my little change as soon as you do that
03:14 est31 pushed
03:14 est31 ok
03:27 Zeno` I dunno about this grey world issue
03:28 Zeno` I wish compiling on Windows was easier heh
03:28 Zeno` not that it's hard... just kind of demented
03:28 technomancy yeah, not much fun to debug
03:28 technomancy for my own projects the policy is always "any windows support is entirely accidental"
03:29 Zeno` :D
03:29 technomancy but that's harder to get away with if you're working on a game
03:30 Zeno` The other problem I have (on Windows) is that bugs other people seem to experience I don't
03:30 Zeno` e.g. there is an issue with main menu apparently but I cannot reproduce it
03:30 technomancy I guess the other problem is it finishes compiling the first half, then it finishes compiling half of what's left, and then it finishes compiling half of the part that remains after that; how is it ever going to actually finish compiling?
03:30 est31 lol
03:31 Zeno` maybe it's the way I run it. I.e. perhaps I have preconceptions that allow me to subconsiously do things that make things work :3
03:31 Zeno` that didn't come out how I meant it to. It made sense when I was thinking it
03:31 est31 the partial sums converge though
03:31 technomancy Zeno`: I heard of a bug in a drawing program that only ever manifested when you draw strokes upward; the developers when testing only ever drew downwards.
03:31 est31 lol
03:32 Zeno` technomancy, I wrote a paradox exactly about what you describe
03:32 * technomancy nods sagely
03:32 Zeno` I call it "Zeno's compilation paradox"
03:34 est31 the trick is the compiler needs for every step one half less time
03:35 Zeno` pfft
03:35 Zeno` as if
03:40 FR^4 joined #minetest-dev
03:40 Zeno` joined #minetest-dev
03:41 Hunterz joined #minetest-dev
03:41 FR^4 joined #minetest-dev
03:52 FR^4 joined #minetest-dev
04:02 FR^4 joined #minetest-dev
04:06 RealBadAngel joined #minetest-dev
04:17 deltib1 joined #minetest-dev
04:32 jin_xi joined #minetest-dev
04:33 Zeno` joined #minetest-dev
04:41 Adimgar joined #minetest-dev
04:44 hmmmm joined #minetest-dev
04:44 est31 joined #minetest-dev
04:47 Adimgar hi, someone awake?
04:47 Adimgar i've some questions about a minetest_game fork i made
04:47 est31 just ask, dont ask to ask :)
04:47 Adimgar lol
04:48 Adimgar may i ask to ask then?
04:48 Adimgar kidding
04:48 est31 no you may not
04:48 est31 also, dont ask to ask to ask
04:48 Adimgar ;(
04:48 Adimgar i talked with kilbith earlier, about translating minetest_game and the engine itself
04:49 Adimgar i made some progress with minetest_game using Intllib from Kaeza
04:50 Adimgar i forked it, made the changes but i made a pull request by accident, i closed it
04:50 est31 so?
04:50 Adimgar it's pretty much complete, it has spanish translations
04:50 Adimgar so at this point i don't know what to do, it's ok with the pull request?
04:52 est31 if you have changes you want to get into minetest, open a pull request
04:52 est31 so what exactly is your question ;)
04:52 Adimgar you've already answered it, thanks
04:52 Adimgar it was if i can go on with the pull
04:53 Adimgar alrigh, it's open again, hope you find it usefull
04:54 Adimgar i'm pretty new on github, i guess everyone can collaborate on making additional language files, right?
04:54 Adimgar i can only translate to spanish
04:55 est31 through pull requests yes
04:55 Adimgar good
04:58 Zeno` We didn't already have a Spanish translation?
04:58 est31 for _game?
04:58 Zeno` oh
04:58 Zeno` cool :)
04:58 est31 of course this mostly only helps in singleplayer
04:58 est31 but a good start
04:59 est31 later on the S() function can be replaced by something that sends the string to translate to the client
04:59 est31 together with information where the string ends and so on
04:59 est31 then the client translates it locally
05:00 est31 its easier I think than recreating the whole infrastructure to send strings to clients based on their language
05:00 est31 of course the client will get a map what translates to what at joining the server
05:01 est31 so that we keep the cool feature of minetest that mods require no client changes
05:02 est31 question remains though about backwards compat
05:02 est31 that *will* require a full change of the structure
05:03 Zeno` maybe
05:03 est31 it can be made optional
05:04 est31 I guess its faster to program a lightweight converter that strips all such things in traffic
05:05 est31 so new clients connect to port A old ones to port B
05:05 est31 and there is a converter B -> A
05:05 est31 acting to A as client
05:05 est31 to B as server
05:05 kahrl est31: maybe add translation support to client and server now, and then when everyone is using new enough versions, add translation in minetest_game
05:06 est31 thats what I meant with "it can be made optional" :)
05:08 Adimgar that was the original idea of it, to try to make the whole thing i18n'd
05:09 Adimgar the S() function in _game it's a little start i believe, but it's better than nothing
05:09 est31 yes
05:09 est31 there are some issues with your patch though
05:09 est31 I'll add them as comments
05:10 Adimgar and like you pointed, it can be improved a lot about reading strings and so
05:10 Adimgar yeah, i expected it, i'm a first timer in all of it :D
05:10 Adimgar thanks for your comments
05:12 est31 quite good for a first timer :)
05:13 kahrl I would suggest _ instead of S
05:13 kahrl _ is a very common alias of gettext
05:13 est31 S is what intllib uses... Im neutral about this
05:14 kahrl hmm ok, I never used intllib
05:14 Zeno` you're not allowed to be neutral. you must have strong, wild, unsubstantiated opinions that cannot be swayed
05:16 kahrl one thing I wonder is: how secure is gettext? is it safe to let it parse potentially malicious .po or .mo files?
05:16 est31 why?
05:17 kahrl because that's how I assume the client would do the translation
05:17 est31 intllib uses simple text files based on lua
05:18 kahrl yeah, but the client has no lua environment
05:18 kahrl not yet, anyway
05:18 kahrl and if one is added there's the same question about security, but for lua
05:18 Zeno` well it kinda does
05:18 Zeno` e.g. mainmenu
05:18 est31 what I have in mind is that the client sends their language in the init, and the server then sends a package with the strings back
05:18 est31 at join
05:19 kahrl a package in what format?
05:19 est31 zipped text file
05:19 est31 https://github.com/Adimgar/minetest_game/​blob/translation/mods/beds/locale/es.txt
05:19 est31 example
05:19 est31 just combined for every mod
05:19 est31 in theory we don't even need lua
05:20 kahrl lol, those formspec artifacts
05:20 kahrl but yeah that should be fine
05:20 est31 ah then we should use gettext yes
05:21 kahrl nah, I don't think gettext is needed
05:21 est31 we can also implement it on our own
05:22 est31 or we wait for clientside lua and do the feature there
05:22 est31 that would require vast API creation though
05:22 est31 it will be faster to code to make it in C
05:23 est31 s/to code//
05:24 kahrl seeing as client side modding is kind of the big vaporware here, I'd prefer not to rely on it ;)
05:24 est31 lol
05:29 hmmmm exactly one person has any real plans for implementing it and zero time
05:30 est31 I have more time the week after next week, I can look into clientside modding if you want.
05:31 hmmmm how do you have so much time
05:31 est31 uni student, and we have a week free. I won't do it all ofc :)
05:32 est31 I mean adding a commit that adds 300 API functions is the wrong approach
05:32 est31 at the start, have some basic things, and then it can be extended
05:32 hmmmm yup, any successful big thing started out as a small thing
05:32 est31 like the current server API started too.
05:32 est31 yea :)
05:33 RealBadAngel joined #minetest-dev
05:35 est31 bye
05:36 hmmmm really i wanted to do the client side modding though
05:36 hmmmm what was i working on last before i left...
05:36 hmmmm ah noise memory optimizations
05:37 hmmmm yeah, that, and then i wanted to optimize a couple of other things like MapBlock::raiseModified()
05:37 hmmmm and then getPlayer() is horrible, why is this not a map
05:37 hmmmm then i wanted to fix the LuaJIT exception wrapper and OS X RUN_IN_PLACE
05:38 hmmmm implement decolsystem finally
05:38 hmmmm and then i can do client side modding
05:40 Zeno` it's C jeopardy time!
05:43 Adimgar and then you can have a cofee :D
05:43 Hunterz joined #minetest-dev
05:47 jin_xi joined #minetest-dev
06:26 Player_2 joined #minetest-dev
06:34 cib0 joined #minetest-dev
07:00 nore joined #minetest-dev
07:13 est31 joined #minetest-dev
07:18 Calinou joined #minetest-dev
07:29 kilbith joined #minetest-dev
08:00 Yepoleb_ joined #minetest-dev
08:01 Krock joined #minetest-dev
08:20 TeTpaAka joined #minetest-dev
08:24 cib0 joined #minetest-dev
08:52 TeTpaAka joined #minetest-dev
09:09 book` joined #minetest-dev
09:10 cib0 joined #minetest-dev
09:14 TeTpaAka joined #minetest-dev
09:16 Darcidride joined #minetest-dev
09:18 TeTpaAka #1307 can be closed now since #2691 got merged.
09:18 ShadowBot https://github.com/minetest/minetest/issues/1307 -- Add minetest.register_on_punchplayer by Bremaweb
09:18 ShadowBot https://github.com/minetest/minetest/issues/2691 -- Add minetest.register_on_punchplayer by TeTpaAka
09:21 est31 done
09:24 TeTpaAka Thanks.
09:25 Megaf joined #minetest-dev
09:26 ElectronLibre joined #minetest-dev
10:30 SudoAptGetPlay joined #minetest-dev
10:59 chchjesus joined #minetest-dev
11:17 Amaz joined #minetest-dev
11:18 proller joined #minetest-dev
11:35 Taoki joined #minetest-dev
11:38 kilbith joined #minetest-dev
11:47 Wayward_Tab joined #minetest-dev
11:55 Wayward_Tab joined #minetest-dev
11:57 cib0 joined #minetest-dev
12:30 OldCoder joined #minetest-dev
12:48 prozacgod joined #minetest-dev
13:28 Wayward_One joined #minetest-dev
13:35 hmmmm joined #minetest-dev
13:40 selat joined #minetest-dev
13:51 proller joined #minetest-dev
14:34 cib0 joined #minetest-dev
14:49 proller joined #minetest-dev
14:50 EvergreenTree joined #minetest-dev
14:51 kilbith joined #minetest-dev
15:00 Hunterz joined #minetest-dev
15:07 EvergreenTree joined #minetest-dev
15:12 younishd joined #minetest-dev
16:25 Adimgar joined #minetest-dev
16:30 rubenwardy joined #minetest-dev
16:39 Hunterz joined #minetest-dev
16:55 MinetestForFun joined #minetest-dev
17:00 TheWild joined #minetest-dev
17:30 cib0 joined #minetest-dev
17:30 blaze joined #minetest-dev
17:32 hmmmm i'd like to make a general statement about adding apis for very mundane things like this https://github.com/minetest/minetest/commi​t/86a963caca9604ad57904e9acd9bef7c46ca47d8
17:34 hmmmm these are such specific functions to accomplish such a specific task.  the problem is that when we go back to expand upon it, these legacy apis need to stay there for reverse compatibility.  this is how things get bloated.
17:34 hmmmm i made the same mistake with the schematic API
17:35 hmmmm in fact now that i look at it, that patch bumps the protocol version
17:38 rubenwardy That would be a useful feature for my CityVsCity subgame. But it could tie in with a new HUD framework.
17:38 hmmmm don't get me wrong
17:39 hmmmm i think the implementation of that feature is spot on
17:39 hmmmm it's just that the feature itself is sort of questionable
17:48 rubenwardy It's useful for teams etc
17:48 rubenwardy but better solved by better hud
17:49 rubenwardy and for admins
17:56 proller joined #minetest-dev
18:00 TheWild why there is set_eye_offset, but no get_eye_offset?
18:13 hmmmm because frankly the api sucks
18:13 hmmmm it wasn't well thought out.  people just added what they needed
18:13 hmmmm it would be really fantastic if we could draw a line in the sand and break all previous mod compatibility and come up with a very good api
18:17 celeron55 well, api design is hard
18:17 Taoki joined #minetest-dev
18:17 celeron55 you might get it right at one time, but then it's going to degrade again
18:19 celeron55 i think a process of introducing experimental apis and then replacing them with proper ones when people actually know how and what for the api is actually used would be ideal
18:19 celeron55 but that isn't easy either
18:19 celeron55 when you have a network layer in between and whatnot
18:20 hmmmm too bad bkvl never happened
18:21 celeron55 it's very doable if someone actually dedicates to it in the long term, but that always kind of takes away from other progress
18:22 celeron55 there's not enough apparent experience and fame to be gained from being someone that maintains it
18:24 hmmmm heh.  i understand the 'lack of fame' part of it.  most users don't really think i work on anything at all
18:24 celeron55 i think everyone will agree that experience and fame is what people are compensated for their work with in this community
18:27 celeron55 hmm
18:28 hmmmm dunno.  i first started working on things to scratch my own itch, and then i continued on because it was interesting and now i feel obligated to the rest of the users to deliver good, finished product
18:28 hmmmm s
18:28 celeron55 i wonder if it would make a significant difference if i would start blogging or otherwise writing about things going on in minetest
18:28 hmmmm probably not
18:29 celeron55 with the intention of making the active people known that don't really do it by themselves
18:29 hmmmm instead of writing about it, make it happen :)  why don't you work on minetest anymore?  you're very much needed in the audiovisual department
18:30 celeron55 i have said this plenty of times already but i can repeat it for this discussion: i don't develop minetest because i don't use it anymore
18:30 celeron55 it's much more natural to work on things that i personally have use for
18:31 hmmmm okay then
18:31 hmmmm i haven't played minetest for years
18:31 celeron55 this includes the fact that back in the day there was a community of finnish minetest users that i played it with
18:32 celeron55 but that community hasn't existed in a long time
18:33 celeron55 i don't really know why and maybe i should figure it out for the benefit of minetest itself, but that's how it is anyway
18:34 hmmmm minecraft isn't popular anymore.  it was a fad and it's over now
18:34 TeTpaAka joined #minetest-dev
18:36 TeTpaAka I'm sorry if my api function isn't good. What should I do better next time?
18:37 hmmmm you personally couldn't
18:38 TeTpaAka What would be needed to make the overall API better?
18:38 hmmmm what you wanted is for the color to be different
18:38 hmmmm you coded it, rather well, you didn't make any mistakes, it's 100% compliant with the code style
18:38 hmmmm it seems pretty solid
18:40 hmmmm this is the fundamental problem with the open source "scratch your own itch" model and designing a coherent standard that works for everybody's use case great
18:40 Calinou use a consortium
18:40 hmmmm i don't like using this word, but i wonder if it would help to make a committee
18:40 Calinou bloat and bureaucracy everywhere :)
18:41 hmmmm yeah, consortium... committee
18:41 Calinou celeron55 is strongly opposed to this
18:41 hmmmm of course
18:41 Calinou he always says people should not design stuff in a commitee
18:41 hmmmm as would anybody
18:42 hmmmm i'm opposed to the idea, but i think it would be necessary in order to keep standardized things coherent
18:42 hmmmm i like the idea of experimental APIs though
18:42 hmmmm we can expand upon the interface or improve it in various ways before it becomes something we need reverse compatibility for
18:43 TeTpaAka Maybe we should revert my pull so I can create an API that would be better forward compatible?
18:43 hmmmm ehh
18:43 TeTpaAka I mean, adding a table that can take more stuff than just the color.
18:44 hmmmm that would be an improvement, make it like core.set_nametag_attributes() or something
18:44 hmmmm but it's not the ideal solution
18:44 TeTpaAka In the future, there might be options to change the size, font, text...
18:44 hmmmm right
18:44 hmmmm if you think about it, this is something client-side
18:45 hmmmm i can't hold it off any longer.  client side modding needs to come right away
18:45 hmmmm every time we add one of these "modify some visual detail on the client side" APIs, that forces me to add a backwards compatibility RPC implementation
18:48 hmmmm maybe i can make all ObjectRef APIs use a generic RPC translation mechanism when player != localplayer
18:49 hmmmm yeah that'll work.
18:49 hmmmm TeTpaAka, sounds good:  change get/set_nametag_color to get/set_nametag_attribute(), have it take a table, and "color" is just one of the attributes
18:50 hmmmm we should be able to seamlessly integrate this with client side modding when the time comes, if done correctly
18:50 TeTpaAka How would I do the network code? Do I have to add a field version, so I can tell how many arguments are transmittet?
18:51 hmmmm actually, I'm not sure
18:52 hmmmm i think the way activeobject messages are implemented right now make things unnecessarily difficult
18:52 hmmmm they shouldn't be positionally based
18:53 hmmmm we need to do some serious thinking about forward/backward compatibility on the network.  i had an idea to transform the protocol into something similar to protobufs but more efficient and completely reverse compatible with the current protocol
18:54 hmmmm maybe it's time to implement it
19:13 Adimgar joined #minetest-dev
19:13 Hijiri joined #minetest-dev
19:31 cib joined #minetest-dev
19:43 proller joined #minetest-dev
19:47 jin_xi joined #minetest-dev
19:53 TeTpaAka #2705 <- is this better?
19:53 ShadowBot https://github.com/minetest/minetest/issues/2705 -- Replace get/set_nametag_color by get/set_nametag_attributes by TeTpaAka
20:01 hmmmm yea
20:08 Miner_48er joined #minetest-dev
20:57 est31 joined #minetest-dev
20:58 est31 minetest is far too small for a comitee
20:58 est31 such things are to make things slow such that communities don't break apart
20:58 est31 as a comparison, coreboot just got a comitee
20:59 crazyR joined #minetest-dev
20:59 crazyR joined #minetest-dev
20:59 est31 and they are much larger :)
21:00 est31 and they have to deal with multiple entities who invest money like hw manufacturers
21:00 est31 those need securities
21:00 est31 either way, about network.
21:00 est31 protobufs sound nice
21:01 est31 now to client modding (I'm switching fast :p), what is your imagined design hmmmm ?
21:01 hmmmm i wanted to do that
21:01 hmmmm :/
21:01 est31 yes you can
21:01 est31 please do it I dont want to take it away
21:02 est31 just want to talk bout design :)
21:02 hmmmm oh
21:02 hmmmm didn't i write on the dev mailing list
21:02 est31 ah
21:02 est31 lemme see
21:02 hmmmm i wanted to have 7 main categories
21:03 hmmmm HUD, the new UI toolkit, a formspec translation layer, sounds, client visual effects, shaders
21:03 hmmmm and something else
21:04 est31 the only message on the list I can see is about envlock
21:04 est31 and your test mail ofc
21:05 hmmmm ahhh
21:05 hmmmm maybe i didn't put it up on there then
21:11 ShadowNinja ...handleCommand_NodeDef *only* involves about 13 string copies...
21:12 est31 prime :D
21:12 ShadowNinja 7 or so actually, std::strings, std::stringstreams, and a Buffer.
21:13 ShadowNinja When you could just use pkt->readLongString() and have 2 or 3 copies.
21:13 ShadowNinja (std::stringstream is overrated, it makes too many copies)
21:13 est31 my guess is that code existed before networkpacket
21:14 crazyR joined #minetest-dev
21:17 ShadowNinja Yes, I've found at least two instances where the entire packet's loaded into an istringstream though, and I'm only halfway through the client packet handlers.
21:17 RealBadAngel joined #minetest-dev
21:17 RealBadAngel hi
21:17 RealBadAngel ShadowNinja, here?
21:17 ShadowNinja RealBadAngel: Yep.
21:18 RealBadAngel im stuck and need some help, do you have some spare time? :)
21:19 ShadowNinja RealBadAngel: A bit, yes.
21:30 RealBadAngel ShadowNinja, got my messages on priv?
21:31 luizrpgluiz joined #minetest-dev
21:31 ShadowNinja RealBadAngel: Now I did. :-)
21:40 rickmcfarley joined #minetest-dev
21:45 hmmmm ShadowNinja, are you sure it's a smart idea to have CHECK_POS_* on read_ functions?
21:45 hmmmm this is not consistent with read_aabb3f
21:46 ShadowNinja hmmmm: Maybe not, but the check was there before so I didn't change it.
21:47 hmmmm fixing all this stuff is something that should be done when we start to unify and clean up the api
21:47 hmmmm consistency is key for a good interface, and none of this is consistent at all
21:49 hmmmm speaking of which, can you take a look at read_object_properties(), the part where it reads colors?
21:49 hmmmm is that correct at all?  readARGB8() expects a table, not a string... why is it checking lua_isstring()...
22:13 ShadowNinja hmmmm: No, it isn't correct.  It was probably supposed to be parseColorString or similar.
22:14 hmmmm nobody complained about it not working
22:15 ShadowNinja Does anything actually use colors?
22:15 hmmmm no idea.
22:15 ShadowNinja Docs say it depends on the visual, but all visuals use textures, not colors AFAIK.
22:15 hmmmm i'm gonna make a generalized read_color() though
22:18 est31 hmmmm got sign comparison warnings again
22:18 hmmmm est31, instead of fixing individual sign comparisons, why don't you let me find the appropriate compiler switch for clang, enable that, and then i'll fix them all in one go.
22:19 est31 you mean I enable it in the cmake script, or just tell it to you?
22:20 est31 hmmmm, ^
22:24 Player_2 joined #minetest-dev
22:24 hmmmm i'm working on it
22:29 luizrpgluiz left #minetest-dev
22:29 est31 nice
22:35 hmmmm erm
22:35 hmmmm just curious
22:35 hmmmm how the hell does minetest compile without irrlicht??
22:36 hmmmm video::SColor is an irrlicht dependency and it's used throughout the SAPI
22:39 ShadowNinja hmmmm: SColor is defined inline in the header.
22:40 ShadowNinja The headers are required but not the lib.
22:40 hmmmm right, so it is required
22:40 hmmmm do we really want this? :/
22:40 proller joined #minetest-dev
22:40 ShadowNinja That's very fragile though, Irrlicht could move some functions and create a dependency on the lib for the server.
22:41 hmmmm exactly
22:41 cib0 joined #minetest-dev
22:42 ShadowNinja We'd need to replace {u,s}{8,16,32,64} (which I've done in one of my brances) core::vector{2,3}d, SColor, probably more.
22:42 hmmmm that's not really too problematic though
22:43 ShadowNinja It would make it slightly easier to switch to something like Ogre3d, if we ever tried that, though.
22:43 hmmmm it'd just consist of copying over those structures into our own file, renaming it slightly, and then running a global find/replace
22:43 ShadowNinja The u8, etc typedefs should use the standard uint8_t, etc typedefs though.
22:44 ShadowNinja https://github.com/ShadowNinja/mi​netest/commits/standard-int-types
22:45 ShadowNinja ^ That uses some using directives to make irr::* namespaces imported by int_types.h, to prevent a bunch of fixing everywhere.
22:46 ShadowNinja Eh, or not.  I guess I fixed that.
22:47 ShadowNinja That might not compile actually.
22:47 ShadowNinja I might not have pushed my latest version.
22:48 est31 u8 is easier to write
22:50 ShadowNinja est31: Yes, the typedef is fine, it just should be changed to a typedef for the standard version.
22:51 est31 yes thats ok
23:00 Megaf Hi, anyone working on #2700 ?
23:00 ShadowBot https://github.com/minetest/minetest/issues/2700 -- [Feature Request] Add command to display the current time.
23:01 est31 I'm right now finalizing packets for protocol v25
23:22 est31 what do you say about this: https://github.com/est31/minetest/commit/​4b6c559a5159c597fbb249374392163f055063f4
23:23 est31 that comit actually enables srp
23:25 est31 hmmmm? ^
23:26 hmmmm what does the value (u16)0 do?
23:26 est31 its a byteflag for compression modes
23:26 hmmmm how would i know that from looking at that function
23:26 est31 right now it isn't used
23:27 hmmmm est, no offense, but you need to work on making your code less cryptic
23:27 est31 it is documented in networkprotocol.h
23:27 est31 ill add a comment
23:27 Hijiri joined #minetest-dev
23:27 hmmmm instead of a comment why not something like
23:27 hmmmm static const current_compression_modes = 0;
23:28 hmmmm static const u16 rather
23:28 est31 yes
23:28 hmmmm compression_modes rather
23:28 hmmmm / Currently, no compression modes have been enabled
23:28 hmmmm or how about...
23:28 hmmmm #define SUPPORTED_COMPRESSION_MODES ((u16)0)
23:29 RealBadAngel ShadowNinja, still here?
23:29 hmmmm I need to still look at the packet structures
23:29 hmmmm BBL
23:29 est31 of srp itself?
23:29 ShadowNinja RealBadAngel: Yes, but working on NetworkPacket.
23:29 RealBadAngel i think i found the reason
23:30 RealBadAngel somehow direct writing to objects in threads is not allowed
23:31 Taoki RealBadAngel: Any news on the water shader and hardware lighting? And hi, since it's been a while ^^
23:31 RealBadAngel v2s16 table[1024]; table[10].X = 255;
23:32 RealBadAngel ^^ this code crashes the thread
23:32 ShadowNinja RealBadAngel: Er, no, you're definitely allowed to write to memory from threads.
23:32 est31 updated
23:32 est31 https://github.com/est31/minetest/commits/master
23:32 RealBadAngel hi Taoki, i will back to shaders coding, now im fighting with minimapper
23:32 Taoki ok
23:33 RealBadAngel ShadowNinja, definitely youre not allowed to, see what code crashes mapblock mesh
23:33 ShadowNinja RealBadAngel: That's really bizzare are you sure that's the exact code that you're using and that's the part that's crashing?  Also, you mean SIGSEGV, right?
23:34 RealBadAngel lemme gdb it
23:34 ShadowNinja RealBadAngel: Everything is a thread, even the main thread is a thread.  Mutexes wouldn't be needed if only one thread could access main memory.
23:35 ShadowNinja ... handleCommand_HudAdd creates an istring stream from the packet (two copies), and doesn't use it.
23:36 RealBadAngel got same backtrace as before
23:43 paramat joined #minetest-dev
23:52 paramat hi hmmmm, i'm getting a segfault (with nothing in verbose debug.txt) when trying to serialize a lua table schematic to mts using file format v4. my code https://gist.github.com/pa​ramat/ec390a34af516d418faf
23:55 rickmcfarley joined #minetest-dev
23:58 Taoki Anyone know if there's a reason why mgv6 biomes aren't enabled by default? I'm trying them out myself, and they look really awesome! Also not experiencing any bugs. Don't see any reason for them not to.
23:59 paramat you mean snowbiomes?
23:59 Taoki snow, yes... sorry about that >_>

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