Minetest logo

IRC log for #minetest-dev, 2014-02-17

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

All times shown according to UTC.

Time Nick Message
00:10 rsiska joined #minetest-dev
00:36 ImQ009 joined #minetest-dev
01:02 hmmmm joined #minetest-dev
02:20 EvergreenTree joined #minetest-dev
02:36 Exio4 joined #minetest-dev
04:32 mrtux joined #minetest-dev
05:34 Miner_48er joined #minetest-dev
05:59 grrk-bzzt joined #minetest-dev
06:01 damiel joined #minetest-dev
06:13 ImQ009 joined #minetest-dev
06:16 xyz sfan5: it seems there's no real need to use crystax ndk
06:16 xyz the official one should work just as fine  these days
06:20 darkrose joined #minetest-dev
06:25 sfan5 xyz: let me try compiling everything using the official one
06:49 RealBadAngel looks like im ready with #1117
06:49 ShadowBot https://github.com/minetest/minetest/issues/1117 -- Normal maps generation on the fly. by RealBadAngel
06:49 RealBadAngel any opinions?
06:50 rsiska joined #minetest-dev
07:00 Gethiox joined #minetest-dev
07:57 Gethiox joined #minetest-dev
08:03 xyz sfan5: even if it doesn't work the preferred approach should be to fix it
08:03 xyz crystax ndk seems a bit ded
08:06 Robby joined #minetest-dev
08:29 xyz|imaginary joined #minetest-dev
08:59 darkrose joined #minetest-dev
08:59 darkrose joined #minetest-dev
09:02 Exio4 joined #minetest-dev
09:32 Weedy_lappy joined #minetest-dev
10:05 casimir joined #minetest-dev
10:13 tomreyn joined #minetest-dev
10:27 ImQ009 joined #minetest-dev
10:52 PilzAdam joined #minetest-dev
10:58 Jordach joined #minetest-dev
11:28 grrk-bzzt joined #minetest-dev
11:48 daspork joined #minetest-dev
12:39 ImQ009 joined #minetest-dev
12:45 xyz sfan5: so how's it?
12:47 xyz > Changed the sys/cdefs.h header so that __WCHAR_TYPE__ is 32-bit for API levels less than 9, which means that wchat_t is 32-bit for all API levels. To restore the previous behavior, define the _WCHAR_IS_8BIT boolean variable. (Issue 57267)
12:47 xyz > from june 2013 original r9 NDK
12:50 tomreyn joined #minetest-dev
13:46 daspork re: register_node. When using the nodebox type, is the geometry you specify via node_box:fixed in any way modifiable at runtime or is that what the identifier `fixed` is suppose to represent?
13:46 daspork If there is documentation on this I would love to be pointed in that direction.
13:50 werwerwer joined #minetest-dev
13:56 proller joined #minetest-dev
14:02 xyz no you can't modify it
14:02 xyz you can't modify other properties either, they are sent to clients once
14:04 daspork k. Was making some auto-connect wiring. The connection part is just visual so I wanted to avoid making tons of nodes and doing swapping.
14:05 xyz this is the only way now
14:06 xyz some people were working on a way to redefine node properties based on metadata
14:06 xyz but this was never fully done
14:06 RealBadAngel nore is workin on it atm
14:07 hmmmm joined #minetest-dev
14:07 RealBadAngel and about wireing there are already several mods that does it
14:07 daspork Cool. When you mention that its sent to the client only once I am less sad about it, and understand.
14:07 RealBadAngel pipeworks, mesecons, technic
14:08 RealBadAngel all of them have autoconnecting routines
14:08 daspork Yeah I looked through those. technic is your RealBadAngel?
14:08 RealBadAngel kinda, its a teamwork
14:10 RealBadAngel also autoconnecting of wires could be done as drawtype in the engine
14:10 RealBadAngel that could save lotsa resources for more advanced mods
14:10 daspork I have used irrlicht a lot, maybe I should look at it.
14:11 RealBadAngel in fact it isnt that hard to code, there are already several drawtypes that does that
14:11 RealBadAngel fences, framed glass for example
14:13 RealBadAngel check out content_mapblock.cpp on drawtypes
14:15 RealBadAngel hmmmm, here?
14:15 daspork yea not a huge deal. just filling sime irrlicht containers
14:16 RealBadAngel in fact autoconnecting in core should be way easier than in lua
14:16 daspork When I saw it used cmake and irrlicht I thought "wow something thats getting mature that I might actually be able to help on...."
14:19 daspork So the question is how generic should the drawtype be to be included. If it was very specific like connecting in six directions and abme to be modified with metadata its a small hack I think
14:19 RealBadAngel i was thinkin on having defined only connections from center to inner sides of the node
14:20 daspork The server should not care as I see it. Its only how its drawn to the user so.
14:20 RealBadAngel and rotate them
14:20 RealBadAngel then add them to form visual connections
14:21 iqualfragile joined #minetest-dev
14:22 RealBadAngel 2 situations i can think of are climbable wires, that would need 2 nodeboxes
14:22 RealBadAngel and 2nd pipelike that would need only one nodebox
14:24 RealBadAngel well, wires would need also just one, from center of the wall to the edge
14:24 RealBadAngel pipes on the other hand from center of the cube to the opposite wall
14:25 daspork Yeah thats a good way to do it. I was kinda doing the IC2 thing and having a middle that gets drawn when nothing is connecting then the connections from the midde to the edges http://i.imgur.com/tGyuhEq.png
14:26 RealBadAngel ah, to avoid overlay glitches (flickering textures) there should be defined junction and rest of the wire
14:27 RealBadAngel so like on the picture, center part and the one that will go from it further
14:29 RealBadAngel in case of single wire going through node junction will remain same size, when 2 or more junction could be scaled up to show the connections
14:32 daspork and it looks like you register the type in script/cpp_api/s_node.cpp and define the identifier  in nodedef.h?
14:42 RealBadAngel well, lemme find my own commit that added glasslike_framed drawtype
14:45 RealBadAngel nodedef.cpp + h, content_mapnode.cpp and that script file
14:47 RealBadAngel https://github.com/minetest/minetest/commi​t/625a4c2e662f6b69b73a2a828d1b08d72e53ff73
14:48 daspork Thanks
14:53 RealBadAngel hmmmm, http://i.imgur.com/7RdorjD.png look at FPS :) ive made it way faster
15:01 sfan5 xyz: haven't gotten to try yet
15:02 sfan5 +a chance
15:05 RealBadAngel sfan5, have you tried #1117 ?
15:05 ShadowBot https://github.com/minetest/minetest/issues/1117 -- Normal maps generation on the fly. by RealBadAngel
15:06 sfan5 not yet
15:06 RealBadAngel its ready to be pushed, that bumpy effect works great with every texture pack
15:07 RealBadAngel with default it looks even better than your bumpy tp ;)
15:08 RealBadAngel http://i.imgur.com/PFYEr1i.jpg
15:08 RealBadAngel ^^ 128x borderlands tp
15:17 sfan5 <RealBadAngel> with default it looks even better than your bumpy tp ;)
15:17 sfan5 that could be because I just use flat textures and you use actual bumpmaps
15:17 RealBadAngel yeah
15:18 daspork RealBadAngel: I assume the usage of glasslike framed is setting the drawtype and tiles = {"frame.png", "glass.png"}?
15:18 RealBadAngel daspork, yes
15:18 RealBadAngel https://github.com/minetest-technic/​framedglass/blob/master/init.lua#L39
15:19 sfan5 RealBadAngel: heres my code in case you want to take a look: https://gist.github.com/sfan5/9021327
15:23 proller joined #minetest-dev
15:23 PilzAdam joined #minetest-dev
15:35 daspork Personally I don't care about or want round pipes/wires so I think im just gonna do this real simple. Allow for a thickness and textures maybe.
15:37 daspork I guess it would also be good to allow it to stick to blocks as an option for something like the red wire in one of those MC mods.
15:37 proller joined #minetest-dev
15:37 daspork ^ instead of being centered
15:40 RealBadAngel daspork, if you will read the nodebox def from node definition it will not matter its round, flat or whatever
15:41 RealBadAngel let the modder decide on the shape
15:42 daspork I wasnt sure if I should mess with NodeBox. I'd have to think about that. I do see how that drawtype works though.
15:46 RealBadAngel if you let the modder define 2 things, a junction and a piece of wire, it will become universal for all kind of wires
15:47 RealBadAngel engine code will have just to rotate them and calculate connections
15:49 rubenwardy joined #minetest-dev
15:49 daspork looking at it I can define a drawtype and still use the exising nodebox struct and methods like: https://github.com/minetest/minetes​t/blob/master/src/mapnode.cpp#L169  to get the geometry etc from the mod
15:51 daspork thats a question kinda
15:52 BlockMen joined #minetest-dev
15:56 daspork or maybe just a nodebox.type....
16:03 Exio4 joined #minetest-dev
16:29 rubenwardy left #minetest-dev
16:34 Calinou joined #minetest-dev
16:51 rubenwardy joined #minetest-dev
17:01 sfan5 xyz: compiles fine with standard android ndk, I think we don't need it anymore
17:10 proller joined #minetest-dev
17:11 NakedFury joined #minetest-dev
17:15 troller joined #minetest-dev
17:15 proller joined #minetest-dev
17:40 Zeitgeist_ joined #minetest-dev
17:55 EvergreenTree joined #minetest-dev
17:55 EvergreenTree joined #minetest-dev
18:14 grrk-bzzt joined #minetest-dev
18:36 sfan5 not this again: E/AndroidRuntime( 7146): java.lang.RuntimeException: Unable to start activity ComponentInfo{net.minetest.minet​est/android.app.NativeActivity}: java.lang.IllegalArgumentException: Unable to load native library: /data/app-lib/net.minetest​.minetest-1/libminetest.so
18:36 sfan5 ^ xyz, this is what happens when I compile without crystax
18:37 sfan5 Cannot load library: soinfo_relocate(linker.cpp:976): cannot locate symbol "__mb_sb_limit" referenced by "libminetest.so"...
18:38 sfan5 this probably means we can't use the official ndk yet
18:40 ShadowNinja sfan5: Maybe this is helpfull: http://lists.freebsd.org/pipermail/fr​eebsd-ports/2007-November/045017.html
18:40 sfan5 well
18:40 sfan5 the general problem is that the bionic libc has bad/none locale support
18:58 VanessaE sfan5: isn't that the same error I get when I try to run your, xyz's, and stu's builds on my tabs?
18:58 VanessaE oh wait, compile time.  nevermine.
18:58 VanessaE nevermind, too.
19:02 sfan5 VanessaE: it is
19:02 sfan5 anyway
19:03 sfan5 I have something you can test on your tablet
19:04 VanessaE sure
19:05 sfan5 VanessaE: http://sfan5.duckdns.org/dltest
19:05 sfan5 adb push dltest /data/local/tmp/dltest
19:06 sfan5 then use adb shell, chmod it to 0775 and run it like this ./dltest <path to libminetest.so>
19:07 VanessaE hold.
19:13 VanessaE argh
19:14 VanessaE how do I get this stupid shell to NOT wrap the lines to 80 chars so I can copy&paste the whole line out
19:14 VanessaE ?
19:14 xyz sfan5: you sure you built it with correct headers and stuff?
19:14 VanessaE (that is, the input line)
19:14 sfan5 xyz: yes
19:14 sfan5 absolutly
19:15 sfan5 VanessaE: put http://www.busybox.net/downloads​/binaries/latest/busybox-armv7l on your phone and use 'busybox ash'
19:15 xyz well let me try this then
19:15 VanessaE already have busybox :)
19:16 xyz sfan5: what library asks for this symbol btw?
19:16 sfan5 dunno
19:16 sfan5 I pasted everything dlerror gave me
19:17 sfan5 also here are the changes I made for it to compile: http://pastie.org/8742992
19:17 sfan5 (to minetest [my fork])
19:19 xyz sfan5: right so you're compiling with "make"? can you make clean and pastebin "make VERBOSE=1" output?
19:19 sfan5 sure
19:19 VanessaE sfan5:  http://pastebin.ubuntu.com/6950521/
19:19 VanessaE oops
19:19 VanessaE forgot to finish the command
19:19 VanessaE ignore that
19:20 VanessaE http://pastebin.ubuntu.com/6950524/
19:20 VanessaE that's the right now.
19:20 VanessaE one*
19:20 VanessaE so it ain't that the lib ain't found -- it's missing a component
19:21 sfan5 ah
19:21 VanessaE what moron decided "not found" was equivalent to "is missing a dependency"?????
19:21 sfan5 your libz seems to be too old
19:21 sfan5 dunnno
19:21 sfan5 I hate that error message too
19:21 VanessaE this is Android 4.1
19:21 sfan5 weeeeee-
19:21 sfan5 guess we need to provice even libz statically to have maximum compatibility
19:22 sfan5 provide*
19:22 xyz VanessaE: hey hey, what about my .so?
19:22 VanessaE watch it require 10x more static libs :)
19:22 VanessaE xyz: don't have it handy at the moment :P
19:22 xyz can you install it?
19:22 sfan5 xyz: I have a guess why it does not work: I compiled curl with libcrystax
19:23 sfan5 lemme try without curl
19:23 xyz sfan5: just pastebin make VERBOSE=1 output
19:23 xyz see
19:23 xyz I said it's your fault :P
19:24 VanessaE xyz: sure.
19:24 VanessaE hold a bit.
19:25 sfan5 VanessaE: could you follow http://dev.minetest.net/Android ? (if you do ping me before you do the minetest step)
19:26 VanessaE sfan5: if I can remember to, sure :P
19:26 Taoki joined #minetest-dev
19:27 VanessaE root@android:/ # /data//local/tmp/dltest /data/data/org.freeminer/lib/libFreeminer.so
19:27 VanessaE Success
19:28 xyz yet it doesn't work?
19:28 VanessaE well I'm trying it now
19:28 VanessaE it's copying its data right now
19:28 xyz ah alright
19:28 xyz that doesn't tell anything though, it'll try to load .so later
19:28 sfan5 xyz: http://ix.io/ayU
19:28 VanessaE data copy finished, crash.
19:29 VanessaE E/AndroidRuntime( 1893): java.lang.RuntimeException: Unable to start activity ComponentInfo{org.freeminer/or​g.freeminer.MyNativeActivity}: java.lang.IllegalArgumentException: Unable to find native library: Freeminer
19:29 xyz what about adb lolcat?
19:29 xyz ah
19:29 xyz it says unable to find
19:29 VanessaE but that's *after* it copied its data
19:29 xyz not unable to load
19:29 xyz data copy is implemented in java
19:29 xyz so it's not related really
19:29 VanessaE oh ok.
19:32 xyz ugh
19:33 VanessaE now, what's the difference here between the result I got from your bin versus sfan5's?  his just seems to need a static zlib, but why does yours fall down?
19:33 xyz ok let's debug it
19:35 ShadowNinja VanessaE: You should probably install a libz shared library...
19:35 sfan5 ShadowNinja: she has one, but her's is apprently too old
19:35 sfan5 (or I'm targetting a too-high android version)
19:36 ShadowNinja Oh, OK.  Update then.  :-)
19:36 VanessaE impossible.
19:36 xyz no that just means your build is fcked
19:36 xyz accept it
19:36 VanessaE no updates are available for this device (I checked)
19:36 ShadowNinja It's rooted, right?
19:36 VanessaE yup
19:36 sfan5 xyz: yep, works perfectly http://i.imgur.com/6sjQ9Ln.png
19:36 ShadowNinja Then it should be doable.
19:36 xyz sfan5: did you recompile all deps?
19:37 sfan5 I recompiled those where I used crystax nd
19:37 sfan5 k
19:37 xyz no, you should recompile all deps
19:37 xyz with same compiler version
19:37 sfan5 wtf
19:37 VanessaE ShadowNinja: doable, perhaps, but I doubt there's a >4.1 build available for this thing :P
19:37 sfan5 that won't make any difference
19:37 xyz and after a make clean
19:37 xyz what did you use to compile them before?
19:38 sfan5 normal android ndk
19:38 xyz what version? what compiler?
19:38 sfan5 anyway libpng, libjpeg or libbzip2 won't make a difference with localisation broken
19:38 sfan5 dunno
19:39 sfan5 those libs I compiled myself won't have any effect on this anyway
19:39 xyz ok now you're literally making me try this
19:39 xyz so wait a bit
19:39 VanessaE lol
19:42 sfan5 also why does minetest need libbzip2 for?
19:42 sfan5 -for
19:44 xyz for irrlicht probably, but irrlicht should include its own
19:45 VanessaE aren't mapblocks also compressed?
19:45 sfan5 no
19:45 sfan5 sqlite has no compress by default
19:45 sfan5 compression*
19:46 sfan5 that means you CAN grep for default:dirt in map.sqlite
19:46 VanessaE huh.  ok.
19:46 xyz we use zlib for that
19:47 xyz no shit you can't
19:47 sfan5 the node list is uncompressed IIRC
19:48 xyz ah, you're probably right
19:48 sfan5 stefan@stefan-pc:~/minetest/worlds/test$ hd map.sqlite | grep default:dirt | head
19:48 sfan5 00002b90  0c 64 65 66 61 75 6c 74  3a 64 69 72 74 00 01 00  |.default:dirt...|
19:48 sfan5 000053f0  0c 64 65 66 61 75 6c 74  3a 64 69 72 74 0a 00 00  |.default:dirt...|
19:48 sfan5 [...]
19:54 rsiska joined #minetest-dev
20:04 Miner_48er joined #minetest-dev
20:05 EvergreenTree VanessaE: Made a better texture for #240 on minetest_game
20:05 ShadowBot EvergreenTree: Error: ProcessTimeoutError: Process #272 (for String.re) aborted due to timeout.
20:05 EvergreenTree lol
20:05 EvergreenTree Shadowbot, you are looking in the wrong git repo
20:05 EvergreenTree https://github.com/minetest/minetest_game/pull/240
20:06 EvergreenTree Anyone willing to merge?
20:09 Megaf joined #minetest-dev
20:10 xyz VanessaE: I have a great idea
20:10 xyz how about you just mail your tablet to some of us?
20:11 Jordach ehm
20:11 xyz ok, to one of us
20:11 Jordach better.
20:11 EvergreenTree *cough* https://github.com/minetest/minetest_game/pull/240 added a better texture
20:11 xyz sfan5: I seem to have the same issue with NDK r9c, g++ 4.6 and 4.8
20:11 xyz that's RETARDED
20:12 xyz sfan5: do you have freetype?
20:12 sfan5 xyz: that's because android's libc sucks
20:12 sfan5 no
20:13 domtron joined #minetest-dev
20:13 xyz why?
20:14 sfan5 why it sucks?
20:14 xyz yup
20:14 sfan5 because it does not support locale stuff completely
20:14 sfan5 and the bits it does support are supported incorrectly
20:14 xyz like what locale stuff?
20:14 sfan5 struct lconv
20:15 sfan5 aka. the thing you had to do to lua/src/llex.c to make it compile
20:15 xyz ah
20:15 xyz that's not a big problem
20:15 sfan5 a libc does not support locale stuff if workarounds like this need to be done
20:16 sfan5 s/if/correctly if/
20:16 sfan5 applying some workarounds for small things like that in llex.c is not a problem
20:16 xyz is that stuff standardized?
20:17 sfan5 but when it comes to bigger things such as real-serious localisation it gets too big to workaround
20:17 xyz or is that some ganooo crap?
20:17 sfan5 dunno
20:17 xyz that could explain
20:17 sfan5 I'm pretty sure locale stuff is in POSIX
20:18 sfan5 but don't bet money on that
20:19 sfan5 THERE!
20:21 sfan5 xyz: http://pubs.opengroup.org/onlinepubs/​9699919799/functions/localeconv.html lconv->decimal_point is POSIX standard
20:21 sfan5 because android didn't have it we had to comment it out
20:22 xyz right, found it
20:24 xyz VanessaE: sorry, for now I'm too lazy to debug this over IRC
20:24 EvergreenTree Sorry if I'm breaking discussion, but does anyone want to test https://github.com/minetes​t/minetest_game/pull/240?
20:24 EvergreenTree *https://github.com/minetest/minetest_game/pull/240
20:24 xyz but what I think we should do, is:
20:24 * sfan5 haz good idea
20:24 sfan5 use static libc
20:25 xyz (part of NativeActivity.java that fails https://gist.github.com/xyzz/c1b0991ad9b809adbdf8)
20:25 xyz so we just need to get ai.applicationInfo.nativeLibraryDir and System.mapLibraryName(libname) outputs from you
20:25 xyz sfan5: what compiler do you use to compile freetype btw?
20:26 sfan5 I don't have freetype compiled
20:26 xyz ah
20:26 xyz well
20:26 xyz BE FUCKING PREPARED
20:26 sfan5 ?
20:27 sfan5 are you wanting to tell me that freetype is a pain in the ass to compile?
20:30 xyz no it's pretty easy
20:30 xyz but if you use crystax's gcc 4.6 and compile for armeabi
20:30 xyz it'll produce invalid output
20:31 xyz java.lang.UnsatisfiedLinkError: dlopen failed: cannot locate symbol "__gnu_thumb1_case_uqi" referenced by "libapplication.so"...
20:31 xyz you see
20:31 xyz there's a big switch in cffgload.c
20:31 xyz like, really big
20:32 xyz or maybe you won't have this, dunno
20:32 EvergreenTree Will someone please look at this?   And the reason I'm repeating is because No one is paying any attention to me
20:32 EvergreenTree https://github.com/minetest/minetest_game/pull/240
20:32 xyz so have fun
20:33 xyz but that's not too bad
20:33 xyz I mean, I had as from crystax ndk segfault on some files
20:34 xyz [02:02:28] <xyz> https://gist.github.com/xyzz/ba3f390​80ce66e8a14d6/raw/5395140ad36ea1fdef​3a9023dd1c9d0c8691b567/gistfile1.txt
20:34 xyz [02:02:46] <xyz> and this is when you realize
20:34 xyz [02:02:48] <xyz> http://cdn.memegenerator.net​/instances/500x/34064327.jpg
20:35 xyz so it really could be a very good idea to switch away from it
20:35 xyz because it's fucking ded
20:35 sfan5 ok, how about a static libc?
20:36 sfan5 freetype compiled, let me try to use it
20:36 domtron joined #minetest-dev
20:38 sfan5 oh awsome!
20:38 xyz why'd you want to use static libc?
20:38 sfan5 because we can then use a libc that supports localisation properly
20:38 xyz (which is bionic)
20:38 xyz it's just a one line change
20:38 EvergreenTree It appears that attention is quite difficult to get attention here.   Any tips on doing so?  :P
20:39 sfan5 apparently I can't tell cmake where to find freetype
20:39 sfan5 xyz: one line change? which line?
20:39 xyz to Lua? you already made that one
20:39 sfan5 <sfan5> but when it comes to bigger things such as real-serious localisation it gets too big to workaround
20:40 xyz I don't get the problem
20:40 sfan5 ok, fix the broken mainemenu for me then
20:40 sfan5 in one line please
20:40 xyz how's that related to glibc/bionic issue?
20:41 sfan5 the crystax ndk adds the stuff bionic does not have (proper locale support)
20:41 xyz proper locale support? yet we still have to pastch that lua file
20:41 sfan5 no we don't
20:42 xyz can you provide a reference on that?
20:42 sfan5 you are doing something wrong
20:42 Gethiox joined #minetest-dev
20:42 sfan5 it compiles just fine for me without patching that line
20:42 sfan5 and the mainmenu works too
20:42 xyz interesting
20:43 xyz ah you probably use crystax/src/locale, right
20:44 sfan5 no
20:44 sfan5 not at compile-time
20:44 sfan5 well yes
20:44 sfan5 the headers of it
20:44 sfan5 lua compiles fine for me with 0 percent changes
20:44 xyz nice
20:45 sfan5 that is 'proof' that crystax adds proper locale support
20:45 xyz yup, figured it out
20:45 sfan5 if we use a static libc we could use one that has full locale support
20:45 sfan5 such as uclibc
20:46 sfan5 the first minetestserver builds I made for android had a static uclibc compiled in
20:46 sfan5 without that the chat did not even work
20:49 xyz so I just checked
20:49 sfan5 to sum it up: it would have the benefit of always being able to use the latest android ndk but having full support for locale stuff and other things bionic misses
20:49 xyz and official NDK 9c provides localeconv as well
20:49 sfan5 but without decimal_point attr
20:49 xyz uh? no, it's here
20:50 sfan5 so it still does not fully comply with the POSIX standard
20:50 sfan5 explain why it fails to compile then?
20:50 xyz sources/android/support/src/locale/localeconv.c
20:50 sfan5 I'm using android-14 as target IIRC
20:50 xyz let's see
20:51 sfan5 android-15*
20:55 sfan5 xyz: http://pastie.org/8743246
20:57 xyz uh
20:57 xyz do you really expect it to work like that?
20:57 xyz it's defined in sources/android/support so you should include this library
20:58 sfan5 posix says it should work like this
20:58 xyz really?
20:58 sfan5 yes
20:58 xyz I don't know so I'll believe you
20:58 sfan5 it says we should include locale.h and not some other file the os provides itself
20:58 xyz indeed
20:59 xyz you include locale.h
20:59 xyz what does posix say about include paths?
20:59 sfan5 dunno
20:59 domtron joined #minetest-dev
20:59 xyz /tmp/ndk-arm/bin/arm-linux-androideabi-gcc -I ~/ndk/android-ndk-r9c/sourc​es/android/support/include/ localeconv_test.c
21:00 xyz so it's still POSIXey right? if you like this word so much
21:00 sfan5 changing all includes because android's libc refuses to adhere posix is pointless
21:00 xyz not all includes
21:00 sfan5 many*
21:00 xyz I didn't change your file at all
21:00 sfan5 you know what I mean
21:00 xyz no I don't
21:00 xyz you just add it to include path
21:00 xyz in one place
21:01 sfan5 won't work
21:01 sfan5 we need to compile android's support library and add it too
21:01 xyz right
21:01 sfan5 I'm pretty sure it won't link/load then
21:01 xyz is that a problem?
21:01 xyz why?
21:01 sfan5 no that is not a problem
21:01 xyz why won't it link or load?
21:02 sfan5 but I don't like how the NDK just ignores any standards
21:02 sfan5 well
21:02 sfan5 it would link and maybe load too
21:02 sfan5 but not work
21:02 sfan5 android's support library defines decimal_point but because bionic does not have it we would access whatever element comes next in the struct
21:03 xyz wat
21:04 sfan5 It makes perfect sense to me
21:04 xyz you just said some crazy shit and now it makes perfect sense to you
21:04 xyz wtf
21:04 sfan5 please tell my what is faulty in my logic
21:06 sfan5 shall I illustrate with an example?
21:06 xyz uh no need to act this smart
21:06 xyz you know how linking works right?
21:06 sfan5 yes I do
21:06 xyz so
21:06 xyz you compile that library which defines lconv, that function and friends
21:06 xyz and then statically link it
21:06 xyz and everything works fine
21:07 xyz in that case
21:07 xyz or what else do you need?
21:08 sfan5 I was explaining you why it would not work if we DIDN'T link to the andrid support library
21:08 sfan5 <sfan5> I'm pretty sure it won't link/load then
21:08 sfan5 <xyz> why?
21:08 xyz [01:02:13] <sfan5> it would link and maybe load too
21:08 xyz [01:02:14] <sfan5> but not work
21:08 xyz <sfan5> I was explaining you why it would not work if we DIDN'T link to the andrid support library
21:08 xyz ah I see
21:09 xyz I guess I missed something here
21:10 sfan5 well
21:10 sfan5 using android's library may be a good idea
21:11 sfan5 but I would prefer to change as little minetest/irrlicht code (includes cmake/make) as possible
21:11 sfan5 changing as little as possible could be achieved with a static libc compiled in (e.g. uclibc)
21:12 sfan5 otherwise we'd need to add these include paths at many places
21:12 sfan5 (this does not mean I'm against using androids' support library)
21:13 xyz you're only supposed to alter Android.mk
21:13 xyz but you use cmake right?
21:14 xyz linking another libc just because of locale stuff seems weird to me
21:15 xyz (I remember, the r word is prohibited in this channel)
21:15 xyz anyway, this doesn't solve the problem with fonts
21:15 sfan5 r word?
21:15 sfan5 yes I use cmake
21:16 xyz which is kinda strange since it appears in both irrlicht's internal and freetype renderers
21:16 sfan5 and I think that is a good idea because future buildsystem changes don't need to be redone in something different
21:16 xyz maybe
21:18 xyz maybe it's another irrlicht bug?
21:19 sfan5 maybe 8)
21:19 sfan5 uh oh compiling uclibc requires linux kernel headers (what did I expect?)
21:19 xyz but I already reported two or three and got ignored so I probably won't bother anymore
21:19 xyz ah
21:19 xyz well, have fun then
21:19 sfan5 android-14 is 3.0.1 according to wikipeida
21:19 xyz like, the one where lighting doesn't work
21:20 sfan5 wikipedia*
21:20 sfan5 oh right
21:20 xyz ugh, you're processing doing (r word) things
21:20 sfan5 how did you fix lightning?
21:20 sfan5 [could you please just rell me what you mean by r word?]
21:21 xyz I just patched it to not use EMT_TRANSPARENT_ALPHA_CHANNEL_REF
21:22 sfan5 I assume the patch is in your repo
21:22 xyz yes
21:22 sfan5 ok, thx
21:22 xyz sure
21:23 xyz and then there's a fix to images being mirrored too
21:23 xyz (why am I always doing all the awful work)
21:24 sfan5 (because you chose to do things before me)
21:24 xyz ah, should've close-sourced irrlicht instead :P
21:24 sfan5 lol
21:26 BrandonReese joined #minetest-dev
21:26 sfan5 android's headers are that bad you can't even compile uclibc
21:27 sfan5 '''' /tmp/android-gcc/sysroot/usr/​include/linux/sysctl.h:25:13: error: expected ':', ',', ';', '}' or '__attribute__' before '*' token ''''
21:28 sfan5 wow, there are even linux headers, I'm surpised
21:28 xyz wait wait wait
21:28 sfan5 ?
21:28 xyz you seriously are building it just because of lconv stuff?
21:28 sfan5 I'm just trying
21:28 sfan5 +to build it
21:28 sfan5 just for fun
21:28 sfan5 to test if it works
21:29 xyz ok
21:29 xyz I bet it won't work
21:29 xyz feel free to waste your time though
21:30 sfan5 it errors directly on the first * in the file (that is not a comment)
21:31 proller joined #minetest-dev
21:31 troller joined #minetest-dev
21:31 sfan5 I wish I coul use clang
21:31 sfan5 it would provide a helpful error
21:31 xyz what's stopping you?
21:32 sfan5 wait
21:32 sfan5 you're right
21:33 sfan5 wat
21:33 sfan5 why is it using gcc
21:33 sfan5 ok, I give up
21:38 xyz sfan5: https://marcansoft.com/transf/meme-gcc.jpg
21:38 Selat joined #minetest-dev
21:39 proller joined #minetest-dev
21:39 xyz that was quick
21:39 xyz you give up on what exactly?
21:51 proller joined #minetest-dev
21:52 blaise joined #minetest-dev
21:52 EvergreenTree https://github.com/minetest/minetest_game/pull/240
21:52 EvergreenTree ^ someone look at it?
21:52 EvergreenTree Aside from VanessaE?
21:53 EvergreenTree I've been told no one cares by PilzAdam so I don't know if it's weven worth the trouble
21:57 EvergreenTree Why does minetest_game even exist if no one pays any attention to it?  I'm just trying to get noticed here
22:02 john_minetest joined #minetest-dev
22:04 xyz dunno
22:05 xyz let's remove the repo then?
22:11 VanessaE EvergreenTree: you have to excuse folks in -dev, when there's an ongoing discussion, sometimes other things will tend to be ignored as offtopic relative to it, until some time later on when the other things can be addressed.
22:13 EvergreenTree Which is basically never it appears
22:13 EvergreenTree *sigh*
22:13 VanessaE naw it's just that minetest_game is harder to get things pushed into than core
22:14 VanessaE because there's only two people who have both the interest AND the commit access to actually push
22:14 EvergreenTree Which is kind of odd if you think about it
22:14 VanessaE (and one of those two is...less than interested these days as he has his own project)
22:15 EvergreenTree You think more caution should be taken when merging something to the engine
22:15 EvergreenTree instead of minetest_game
22:15 EvergreenTree I always thought it would be more difficult to merge to core
22:15 EvergreenTree I guess I was wrong
22:15 EvergreenTree :P
22:16 VanessaE there are more people with both commit access, interest, and the skills to code for stuff that goes to core, that's why it's easier to get work done on it.
22:16 VanessaE s/both//
22:16 PilzAdam also there is that semi-official freeze of minetest_game
22:16 VanessaE that too
22:17 EvergreenTree The reason I don't make pull requests for core is I don't know C
22:17 VanessaE but it's only frozen in so far as adding some major feature is concerned.
22:17 EvergreenTree But I still want to contribute, so I try and do stuff on minetset_game
22:17 proller joined #minetest-dev
22:17 VanessaE PilzAdam: I think the point of the freeze is to prevent mt_game from breaking compatibility with the existing mods' code base and/or making major gameplay changes.
22:17 VanessaE isn't it?
22:18 PilzAdam there are several reasons
22:18 domtron joined #minetest-dev
22:18 VanessaE I don't recall anyone agreeing to an absolute moratorium on incremental improvements to the default game.
22:18 PilzAdam one is that its almost impossible to change stuff without a shitstorm
22:19 PilzAdam then we want to encourage that more games get created, instead of focusing on one
22:19 VanessaE yes, and both of those boil down to backwards compatibility with the existing modders' codebases
22:20 domtron joined #minetest-dev
22:20 EvergreenTree The problem with not focusing on the default game is that any random newbie who comes on isn't going to know how to install a game, and is going to be turned away by the fact that minetest_game iis boring
22:21 EvergreenTree so we are going to be stuck with elitists
22:21 domtron joined #minetest-dev
22:21 VanessaE asdly, he's right :(
22:21 VanessaE sadly, too.
22:21 PilzAdam EvergreenTree, there are 2 ways to fix that: 1) shipping more games with the engine 2) dont ship any games with the engine
22:21 VanessaE so, drought or deluge?
22:22 EvergreenTree Not shipping any games with the engine is even worse then having aboring default game
22:22 PilzAdam how so?
22:22 EvergreenTree -aboring +a +boring
22:22 PilzAdam there could be a nice in-game game store or so
22:22 EvergreenTree The point I just made is the fact that minetest is rarely used as an engine
22:22 PilzAdam or rather, in-menu
22:22 EvergreenTree PilzAdam: Then do that
22:23 VanessaE an in-client "game store" would be a plus
22:23 EvergreenTree If that's what you want, why haven't you done it yet?
22:23 EvergreenTree The problem is that there are barely any games
22:23 PilzAdam I dont have the time currently
22:23 EvergreenTree that aren't deprecated
22:23 EvergreenTree by "you" I mean the core devs
22:24 VanessaE At some point there is a plan to include Nodetopia, right?
22:24 PilzAdam there are 1) eden, 2) Nodetopia 3) minetest_game 4) OCD 5) Wasteland 6) capture the flag 7) c55's dungeon game
22:25 VanessaE And BloatTest ;)
22:25 PilzAdam you finally found a name for your game?
22:25 VanessaE nope
22:25 EvergreenTree I thought wasteland was a mod?
22:25 VanessaE but you figured out which one I was talking about :D
22:25 PilzAdam well, your game is basically modded minetest_game
22:25 VanessaE yup
22:26 PilzAdam while all the other listed games are not based on each other
22:26 VanessaE it's minetest_game modded about as far as you can logically take it
22:26 EvergreenTree The problem with not having a default game is that there isn't anything to go off of
22:26 EvergreenTree there should be a default
22:26 PilzAdam well, I guess capture the flag is based on mt_game
22:26 PilzAdam EvergreenTree, thats exactly my point: there cant be _one_ default
22:26 EvergreenTree Well, It's nice to know that all my efforts on minetest_game have gone to waste
22:27 PilzAdam if there is one default game then youll face the problems we currently have with mt_game
22:27 VanessaE I should continue to tune the one I use on 30005 and release the result as "FrenchVanilla"
22:27 VanessaE :)
22:27 EvergreenTree I guess I shouldn't even bother with developing minetest if this is the response I'm going to get
22:27 VanessaE EvergreenTree: one sec..
22:28 VanessaE PilzAdam: to be fair, mt_game isn't entirely dead - Sam II did just go in.  that part was welcomed by pretty much everyone.  the capes thing was decried by a few people (myself included) because it violated standards, but no one opted to remove it
22:28 VanessaE THAT is the sort of "shitstorm" we need to be wary of.
22:29 VanessaE not because there are changes, but because we need to be acutely aware of what our users have come to expect as "standard practice"
22:30 PilzAdam what about finite lava?
22:30 PilzAdam it didnt break anything
22:30 VanessaE PilzAdam: most people hated that change.  I personally don't care much about it.
22:30 PilzAdam your theory that its all about mod compatibility is wrong
22:31 EvergreenTree To be honest: Is adding an animated furnace front to minetest going to cause a "sh*tstorm"?
22:31 VanessaE ok let's make a guess:
22:31 EvergreenTree It isn't even a change to the gameplay
22:31 VanessaE what exactly do you figure is likely to happen should evergreen's animated furnace go in?
22:32 john_minetest EvergreenTree will get +1 in his commit history.
22:32 john_minetest :D
22:32 VanessaE ok, I'll hate evergreen for about a week until I get around to making an appropriate HDX texture to match ;) but that's about it :)
22:32 EvergreenTree ^ that is what should be considered when something is merged to minetest_game
22:32 PilzAdam when nobody is working on minetest_game (for reasons we just discussed) then also nobody is interested in these little tweaks
22:32 EvergreenTree VanessaE:  lo
22:32 PilzAdam thats why they mostly get ignored
22:32 VanessaE but really, this one's so minor that no one's gonna complain about it
22:32 EvergreenTree *lol
22:33 VanessaE *throws washcloth down on table*
22:33 EvergreenTree I'll bet every texture pack creator is going to be mad at me for a little while
22:33 EvergreenTree :P
22:33 VanessaE All right.  you know what?  give ME commit access and I will push it myself.
22:33 EvergreenTree In fact, I'll have to create a new texture for mt-faithful
22:34 VanessaE (the washcloth was handy because I'm munching on cheese doodles and avoiding orange finger syndrome :P )
22:34 EvergreenTree It isn't that difficult to do
22:34 Jordach VanessaE for minetest_game maintanence
22:34 EvergreenTree No one is really in charge of minetest_game atm
22:34 EvergreenTree Because none of the core devs care about it
22:34 VanessaE ok maybe not, because the second thing I'd do, Jordach, is revert the capes thing :P
22:35 PilzAdam didnt I already suggest VanessaE as minetest_game maintainer a while ago?
22:35 VanessaE and the third thing I'd do is merke Taoki's bugfix to the player model.
22:35 PilzAdam and didnt she refuse to do it?
22:35 EvergreenTree Also, texture pack creators don't even really need to do anything
22:35 domtron is the default "mod" separate from mintest_game? A lot of mods depend on it so it should be improved right?
22:35 VanessaE PilzAdam: you did and I declined - I didn't want the responsibility at the time.
22:35 VanessaE domtron: no, it's part of it, but it could be used separately with few changes.
22:36 EvergreenTree Do you now?   It's not very difficult considering the low traffic on it
22:36 PilzAdam "default" is basically a huge bad design decission
22:36 VanessaE PilzAdam: see what RealTest did - it split "default" into a bunch of sub-mods.
22:36 VanessaE real pain in the ass to find anything in THAT game
22:36 PilzAdam VanessaE, guess what I did in Nodetopia
22:39 EvergreenTree You
22:39 EvergreenTree whoops
22:39 EvergreenTree You're right about that, the name "default" is a bad naming choice
22:40 VanessaE indeed
22:41 VanessaE in retrospect it was a bad choice,
22:41 VanessaE it can be fixed but it'll take time.
22:41 EvergreenTree Yes
22:41 EvergreenTree It would be a huge change considering the amount of stuff that relies on the default name
22:41 VanessaE yesh
22:41 VanessaE yes
22:42 Jordach minetest.register_alias is your friend
22:42 VanessaE that won't help depends.txt
22:42 EvergreenTree yep
22:43 EvergreenTree and it still requires mod creators to update their mods
22:43 VanessaE no need to change the node names (though that's what aliases are for) - the problem is dependency upheaval.
22:43 VanessaE you just can't do that now, not unless you just want to break EVERYTHING.
22:43 VanessaE either that or the engine needs to have a compatibility flag
22:43 EvergreenTree ^ that's the problem
22:44 Jordach VanessaE, my suggested method is to keep the current one, and start over for 0.5.x
22:44 EvergreenTree aliases won't help in some cases
22:44 VanessaE if it sees "default" in depends.txt, it automatically searches for libs x, y, z....
22:44 VanessaE but that's a bandaid
22:45 Jordach why don't we just tear down the parchment and tell them to start over if they don't like it, deal with it (there does need some dictatorship around here)
22:45 EvergreenTree I think it's kind of funny that we were talking about merging #240 30 minutes ago, and now we are talking about changing the name default
22:45 ShadowBot https://github.com/minetest/minetest/issues/240 -- Creative inventory
22:45 VanessaE the problem of a dictatorship is....who's the dictator?
22:45 EvergreenTree ShadowBot: wrong pull
22:45 Jordach VanessaE, which means there does need to be a maintainer of minetest_game
22:46 VanessaE Shadowbot is supposed to refer to minetest_game for ID's < 1000 isn't it?
22:46 ShadowBot VanessaE: Error: Missing ">". You may want to quote your arguments with double quotes in order to prevent extra brackets from being evaluated as nested commands.
22:46 VanessaE shaddup, bot :P
22:46 EvergreenTree lol
22:46 VanessaE EvergreenTree: 240 -> changing default....welcome to -dev.
22:47 VanessaE but it all goes together, it's all on the same subject:  how to properly maintain minetest_game and keep it from becoming stale.
22:48 VanessaE my 30005 server has inexplicably become very popular lately, and it runs a very-lightly-modded mt_game (only enough stuff added to make administration a breeze, plus a couple of things to keep it interesting, e.g. gloopblocks, snowdrift)
22:48 VanessaE and it's because of these tablet clients
22:48 VanessaE so a "simple" game has suddenly become acutely necessary
22:49 VanessaE which means mt_game can't die.  we have to have a simple, default game that is actively maintained.
22:50 Jordach here's the deal: we have two games: minetest_mini and minetest_mega, mega being the new game, while mini takes the traits of the current mt_game
22:50 EvergreenTree We tried the multiple games thing before
22:50 VanessaE Jordach: three - mini (Nodetopia), standard (mt_game), mega (my game, perhaps).
22:50 EvergreenTree it didn't really work
22:50 VanessaE EvergreenTree: no, we tried a shanda navara before.
22:50 domtron joined #minetest-dev
22:50 EvergreenTree Nodetopia?
22:50 EvergreenTree Mini?
22:51 Jordach VanessaE, get the modders involved with your pack first
22:51 VanessaE EvergreenTree: that thing we did before was an abortion :P
22:51 EvergreenTree oh right, I was thinking of eden
22:51 EvergreenTree nodetopia would be perfect for mini
22:51 EvergreenTree VanessaE: true
22:51 PilzAdam EvergreenTree, Nodetopia isnt finished at all
22:51 EvergreenTree By the way, is the new texture better?
22:51 EvergreenTree PilzAdam: didn't say it was
22:51 VanessaE it seems better, yeah
22:51 EvergreenTree but it's still very good for a minimalistic game
22:51 Jordach i meant mega to be built from the current game, but with simple mobs, farming+, MGV7, technic(?!!), more biomes, actual things to accomplish
22:51 VanessaE I have not tried it yet, just looked at the pull.
22:52 domtron joined #minetest-dev
22:52 EvergreenTree mega shouldn't have mobs simply because they are too buggy, but that's another topic
22:52 Jordach and probably some bits inbetween to half cool stuff out of THAT game, and some generic survival inbetween the cement
22:53 VanessaE EvergreenTree: actually mobs are fine if you don't have to deal with a network connection between you and the server process (or you don't care about the bandwidth usage)
22:53 VanessaE simple mobs is a good mod to use for the purpose
22:53 EvergreenTree Speaking of changes to minetest_game, shouldn't lava destroy items when thrown into it?
22:53 VanessaE EvergreenTree: builtin_features does that
22:53 VanessaE (I think technic also has that option)
22:53 EvergreenTree Oh, okay
22:54 Jordach a trash field is more effective for clearing cobble
22:54 Jordach or just a button to clear cobble in general
22:54 VanessaE one of the biggest requests from my users was xyz's sethome mod
22:54 VanessaE so I broke down and installed it
22:54 EvergreenTree *cough*  Trash cans *cough*
22:54 VanessaE so that one should be slated for inclusion in mt_game
22:54 Jordach and no: home decor doesn't fit with the mega idea
22:54 EvergreenTree Maybe as an option?   It is kind of cheaty
22:55 Jordach EvergreenTree, we already have /pulverize
22:55 VanessaE EvergreenTree: not really cheaty - you can already teleport yourself anyway if you're running in singleplayer.
22:55 EvergreenTree no, I'm talkingto VanessaE
22:55 VanessaE /grant singleplayer all  ;  /teleport (some coords)
22:55 EvergreenTree VanessaE: I suppose so
22:55 VanessaE it just gives you a shortcut for it.
22:56 EvergreenTree But the option should just be there for server creators(I suppose you could just remove the code for it)
22:56 EvergreenTree most servers run custom games anyway
22:56 VanessaE for a server operator yeah, you'd just delete the mod if you don't want it.
22:58 EvergreenTree Back to the topic:  Should 240 be merged?
22:58 EvergreenTree The discussion was fun though
22:59 VanessaE I'd say yes, it does nothing to change gameplay, and makes the furnaces look a little better.
23:00 EvergreenTree Anyone else?
23:07 EvergreenTree Anyone else willing to merge?
23:09 EvergreenTree I'm starting to wonder if I accomplished anything with this discussion
23:09 EvergreenTree Because the moment I mention merging something, the discussion stops
23:09 EvergreenTree :P
23:09 VanessaE that's just how this channel is :P
23:09 * VanessaE pokes ShadowNinja
23:10 ShadowNinja Hmmm?
23:10 VanessaE ShadowNinja: can we get a merge/no-merge on mt_game pull #240 ?
23:10 ShadowBot https://github.com/minetest/minetest/issues/240 -- Creative inventory
23:11 VanessaE and can you fix THAT ^^^^^ also? :P
23:11 VanessaE (the bot always quotes the wrong pull when the number is < 1000 )
23:11 ShadowNinja VanessaE: I switched it on the request of other devs.
23:11 VanessaE oh, well it's still wrong :P
23:11 VanessaE https://github.com/minetest/minetest_game/pull/240 <--- that one.
23:11 ShadowNinja And that pull needs one ore indentation level, I haven't checked the textures.
23:12 VanessaE oh yes, I see it.
23:12 VanessaE EvergreenTree: lines 899-908, indent by 1 level.
23:13 VanessaE (I rarely look at code style when I'm checking a pull, unless it just plain burns my eyes :P )
23:13 EvergreenTree oh, right
23:14 OldCoder joined #minetest-dev
23:15 VanessaE ShadowNinja: btw, is it just for me that github no longer shows image changes in some pulls?
23:15 EvergreenTree done
23:16 EvergreenTree Is it good now?
23:16 VanessaE no, you need to push 904-908 over to the right, they should be indented past 903
23:16 VanessaE well AT 903
23:17 VanessaE so just move them over one also.
23:17 EvergreenTree oh right
23:18 EvergreenTree Done
23:19 ShadowNinja Seems good.  I'll push it soon of nobody objects.
23:19 VanessaE I'd move 905-907 over one more level also but this is good enough I think.
23:20 EvergreenTree Okay, I can do that quickly
23:21 EvergreenTree done
23:21 VanessaE that's good.
23:22 ShadowNinja Don't bother toughing it anymore.  I have it prepared locally.
23:22 ShadowNinja touching*
23:22 EvergreenTree okay
23:22 OldCoder joined #minetest-dev
23:23 VanessaE he can't toughen it up any?  well shit.  so much for adding rebar to the mortar in the furnace....
23:23 * VanessaE ducks
23:23 EvergreenTree lol
23:42 * EvergreenTree pokes ShadowNinja
23:44 OldCoder joined #minetest-dev
23:45 EvergreenTree Okay, can someone else merge it?
23:45 EvergreenTree he said he would merge it in 15 minutes
23:45 VanessaE "patience, young padawan."
23:47 EvergreenTree Woohoo star wars references
23:47 ShadowNinja d5fcae0..a5dcf12  master -> master
23:48 VanessaE http://www.youtube.com/watch?v=81LQnWPVqSI
23:49 EvergreenTree I also created a mew branch this time
23:49 EvergreenTree -m +n

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