Minetest logo

IRC log for #minetest-dev, 2013-06-09

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

All times shown according to UTC.

Time Nick Message
01:23 mrtux joined #minetest-dev
01:27 MJaoune joined #minetest-dev
01:31 MNJaoune joined #minetest-dev
01:39 kahrl joined #minetest-dev
02:04 khonkhortisan the simplest ones get merged first https://github.com/minetest/minetest/pull/764
03:07 dexter0 joined #minetest-dev
03:41 hmmmm joined #minetest-dev
04:58 Taoki[mobile] joined #minetest-dev
05:57 dexter0 joined #minetest-dev
07:06 Guest83169 joined #minetest-dev
07:15 darkrose joined #minetest-dev
07:15 darkrose joined #minetest-dev
07:32 dexter00 joined #minetest-dev
07:34 zweipktf1t joined #minetest-dev
07:37 ecube joined #minetest-dev
07:37 thexyz_ joined #minetest-dev
07:38 ssieb joined #minetest-dev
07:40 neko259 joined #minetest-dev
08:05 Calinou joined #minetest-dev
08:26 EdB joined #minetest-dev
08:32 serengeor joined #minetest-dev
08:38 Taoki joined #minetest-dev
09:18 ImQ009 joined #minetest-dev
10:21 Zeg9 joined #minetest-dev
10:29 serengeor joined #minetest-dev
10:56 PilzAdam joined #minetest-dev
14:04 Calinou joined #minetest-dev
14:09 Jordach joined #minetest-dev
14:09 Jordach joined #minetest-dev
14:25 sapier1 joined #minetest-dev
14:27 sapier1 does anyone know why it's impossible to link minetest against mingw libz?
14:29 Exio joined #minetest-dev
14:30 rubenwardy joined #minetest-dev
14:46 sapier1 https://github.com/sapier/minetest/tree/ngmm_minizip I've added libarchive support, as well as fixed build on win32 ... win32 builds and links but doesn't work I guess this is due to linker mess created by non mingw libz version
15:23 Neological joined #minetest-dev
15:34 ShadowNinja sapier1: Why the minizip support?
15:35 sapier1 to support users with installing mods
15:35 sapier1 no I'm not talking about developers but USERS ;-)
15:36 Jordach no more extraction of mods / modpacks
15:36 sapier1 exactly and of course no more wrong folders
15:36 ShadowNinja Hmmm, will it work with the GitHub archives properly?
15:37 sapier1 modmanager included within formspec mainmenu will  ... at least in 99.9%
15:37 sapier1 it's detecting modname after extracting it
15:38 Exio how hard would be to add things like a "info' file in the mods
15:38 sapier1 if there is an register command it'll be able to rename the mod automaticaly
15:39 sapier1 at least as long as modname is within same line as register :-) I didn't want to write a full parser ... but this is not normal code style ... thus yes if a mod writer intentionally writes unsupported code it will fail
15:39 sapier1 exio not that hard but it has to support current mods
15:40 sapier1 but I need someone to fix the win32 build as those users are the ones that really need it ;-)
15:40 Exio yes, but i mean "if it exists; use it; if not guess it"
15:41 sapier1 yes I guess this will be added when moddb is added too as moddb already provides a defined data format
15:42 Exio ah
15:42 sapier1 still I'm unable to link zlib correctly on win32
15:42 kahrl minetest already links with zlib, why does it fail?
15:43 sapier1 it does link but it uses some workarounds and fails on startup
15:43 sapier1 if I try to link against mingw libz it doesn't even link
15:46 sapier i guess irrlicht lib does use different calling convention than mingw ... but I'm not common to win32 linkage problems so I may be completely wrong
15:46 NakedFury joined #minetest-dev
15:48 sapier error messages in mingw are "Warning: resolving __unzGetGlobalInfo by linking to __unzGetGlobalInfo@@8"
15:48 sapier and some similar things
15:48 kahrl check src/serialization.cpp
15:48 sapier am I correctly interpreting @@8 as pic/non-pic error?
15:48 kahrl it has #ifdef _WIN32 #define ZLIB_WINAPI
15:49 sapier I already added this without that it didn't even link
15:49 sapier but it's still wrong
15:50 kahrl are you sure you are defining it before zlib.h is included?
15:50 sapier yes
15:50 kahrl checked all header files that are included before it too?
15:50 hmmmm joined #minetest-dev
15:51 sapier zlib.h is first (non c) header to be included in unzip.c
15:52 kahrl and how do you link with it in cmake?
15:53 sapier yes kahrl it does link but it's linking wrong things :-)
15:54 kahrl can you push your changes somewhere?
15:55 sapier btw how to link curl without beeing able to set curl dll in gui?
15:55 sapier it's already up on the branch I posted
15:55 kahrl oh sorry I was looking at next_gen_main_menu
15:55 sapier no the minizip branch
15:55 kahrl I see it now
15:56 sapier mainmenu won't build on win32 atm I didn't merge the fixes back by now
15:56 dysoco joined #minetest-dev
15:57 kahrl is there any particular reason to use an external libarchive?
15:57 kahrl tar seems simple to enough to just parse it a small routine that calls zlib to decompress
15:57 kahrl in*
15:57 kahrl -to
15:58 sapier libarchive is way to support tar.gz tar.bz2 tar.7z ... it supports zip too but I guess not everyone wants to install libarchive
15:58 sapier but it doesn't link correct without libarchive too
15:58 kahrl personally I'm not sure if I like adding all these dependencies into minetest
15:59 sapier ok plz add libz ;-)
15:59 kahrl libz is already used
15:59 sapier into minetest ;-)
16:00 sapier I'd prefere not to use two different libs but I'm not sure if libarchive license is compatible to minetest
16:01 kahrl is it really important that minetest can extract archives that the user downloaded outside of minetest?
16:01 kahrl I mean if he does that he can install whatever unzip program is needed
16:01 sapier atm this is the only way a user can get a mod ;-)
16:01 sapier unzip can't be used on windows
16:01 kahrl huh?
16:02 sapier and linux users normaly don't need support
16:02 kahrl mingw has unzip and there's also 7zip/winzip/winrar/... and explorer can extract zips too
16:02 sapier I don't want to write support for whatever gui unzippers are out there
16:03 sapier yes but microsoft doesn't give api access to unzip capabilities of explorer
16:03 kahrl err, I'm not sure I catch your drift
16:03 kahrl what does minetest have to do with it
16:04 sapier it's just removing an additional silly error case
16:04 kahrl the wrong directory names?
16:04 sapier wrong directory names wrong folders
16:04 sapier telling a noob where to copy a mod is quite callanging
16:05 kahrl noobs can use the included mod manager
16:05 sapier yes but the modmanager needs to unzip ;-)
16:05 kahrl but only one format, not a hundred
16:06 sapier true but it's no difference to support 1 or hundred formats with libarchive
16:07 sapier but even if we only support one format (zip) linking is done wrong ;-)
16:07 celeron55_ my opinion on the issue is that if it is possible to add support for a single common archive format, no matter what, without additional dependencies, it should be done that way
16:07 celeron55_ and it definitely is
16:07 sapier so once we fix this error we can still decide if we want to support tar.gz ...
16:08 celeron55_ people can easily transition to it and in the future things will be much simpler
16:08 kahrl libarchive seems a bit like "mit Kanonen auf Spatzen geschossen"
16:08 kahrl what is that in english...
16:08 kahrl taking a sledgehammer to crack a nut
16:08 sapier no matter if we want zip only support or not linkage is done wrong on windows in any case ;-)
16:08 PilzAdam shooting with canons at flies?
16:09 celeron55_ if we hint to people that we want to support many formats, then people expect us to do that always in the future, and even expect support for even more formats
16:09 kahrl sapier, I will try to write a tiny tar extract function
16:10 kahrl but probably not before thursday because I'm busy
16:10 sapier ok I'll not include libarchive support in mainmenu ;-P still I need a fix for libz linkage on win32
16:10 celeron55_ linux is cancer in this aspect because it makes dependency management so much easier than other OSes
16:10 sapier kahrl a tar extract function is useless without gz support
16:10 kahrl zlib does g(un)zip, and minetest already links with it
16:11 sapier I don't understand why ppl always reinvent the wheel
16:11 sapier either we only support one format so we can live with zip
16:12 sapier or we should use a well maintained lib to support tar
16:12 sapier libarchive is bsd so maybe we could even include it with minetest but I didn't have a closer look at licenses
16:12 sapier btw minizip (the code i added for zip) seems to be included with zlib 1.2.8
16:13 sapier in contrib folder
16:13 kahrl I don't like adding too much useless code to minetest because it makes compile times horrible
16:13 kahrl and most of libarchive will be useless, such as the compression routines
16:13 sapier so plz help me fix the zlib linkage on windows ;-)
16:13 sapier no libarchive uses zlib
16:14 kahrl I mean the functions that write archive formats
16:14 sapier no they're used to extract a tar
16:14 celeron55_ 8D
16:14 celeron55_ ->
16:14 sapier libarchive doesn't make a difference if you want to write to a file or disk
16:15 BlockMen joined #minetest-dev
16:16 sapier no matter kahrl if you disable libarchive support in my branch minetest still will crash on startup due to linker mess
16:17 kahrl sapier: can you do a make clean; make VERBOSE=1 and paste the results?
16:17 sapier if you can tell me how to get build output on mingw windows console?
16:18 kahrl make VERBOSE=1 >output.txt 2>&1
16:18 sapier on windows?
16:19 kahrl are you not using msys?
16:19 sapier I'll try
16:19 kahrl anyway cmd.exe supports redirection too but I'm not sure about the syntax
16:20 sapier ok seems to work
16:20 sapier but needs some time
16:25 sapier is there a size limit on pastebin? :-)
16:25 kahrl heh
16:25 sapier 75% but already 100kb
16:26 kahrl pastebin.com has 512kb
16:27 kahrl most of the code pastebins probably less
16:27 sapier should be enough
16:30 sapier ok once again this build was crap
16:35 sapier kahrl do you have a win32 build env?
16:35 kahrl no
16:36 BlockMen left #minetest-dev
16:36 sapier :-) I'm not even sure it's not only due to my environment ;-)
16:36 sapier it's working on linux
16:36 sapier both minizip as well as libarchive variant
16:36 kahrl hmm
16:39 Miner_48er joined #minetest-dev
16:39 kahrl I guess it has to do with the link order, since everything else seems to be okay
16:40 sapier grrr clean doesn't work on windows either
16:40 kahrl have you tried putting the target_link_libraries for the ARCHIVE_LIBRARY before the one for ZLIB_LIBRARIES
16:42 sapier kahrl I try to get it working without libarchive first :-)
16:42 sapier it already failes with zlib/minizip source
16:43 sapier but this may be related to msys providing it's own libz implementation
16:59 Jordach i have noticed that PilzAdam has been unfair in the way he allows pull request merging: http://irc.minetest.ru/minetest/2013-06-09#i_3131063
16:59 Jordach http://irc.minetest.ru/minetest/2013-06-09#i_3131165
17:01 Jordach if he likes something in a certain way, it has a greater chance of a merge, but if it has something he hates, or just simply thinks it doesnt fit, he closes and throws it away
17:01 sapier I didn't realize any mobs beeing upsreamed did I miss something?
17:02 Jordach sapier, i am sure that PilzAdam will automatically vouch for Simple Mobs
17:03 sapier lol ok that'd be adding a really really buggy mob implementation
17:03 Jordach (also: if it's his own pull, he'll merge it and say everyone likes it)
17:03 Calinou simple mobs are "buggy and slow"
17:03 sapier no just buggy because fixing the bugs will make them slow
17:04 Jordach no Lua mobs mod will be useful until main routines are actually in core
17:05 Jordach sapier, consider main core routines for mobs
17:05 Jordach and then expose them
17:05 sapier I already did jordach it's just not as easy as it sounds at first
17:06 Jordach sapier, even though AI (eg, movement, attacking, projectiles) still need to be in corfe
17:06 Jordach core*
17:06 sapier to avoid interfering mobs with core they need to be handled in a different thread ... but there's lot of preparations to do so
17:06 sapier problem at moving them to core is you'll always loose configurability atm I try to move those things that are really necessary but keep those who don't
17:08 sapier and still if you try mobf without agressive mobs you'll see there isn't that much load added with lua either
17:08 Jordach sapier, projectile_entity with the same options as a regular entity, with the two extras: on contact - eg, node, player etc
17:08 Jordach and node_damage / player_damage
17:08 sapier don't see how this would improove anything
17:09 sapier atm there are two big cpu hogs environment evaluation and player/mob interaction
17:10 sapier moving first one to core will drop configurability almost completely thus different behaviour would be very difficult to implement
17:10 sapier second one isn't very easy to optimize (but is way more promising)
17:10 sapier still there are some pull requests waiting and optimizing mobs to and they are much more simple
17:10 Jordach sapier, most mobs on their own usually just walk about randomly
17:11 sapier jordach that's not true
17:11 sapier once a player is near them they are supposed to react
17:11 Jordach yes
17:11 Jordach a chase param would be useful
17:11 Jordach "player", would be nearest player under x,y,z
17:12 sapier chase isn't enough too as some should run away ;-)
17:12 Jordach flee = "player"
17:12 sapier while some should avoid this surface others that
17:12 sapier you see it's much more complicated ... it's easy to get the 90% cases (that's what simple mobs does) ... the other 10% are hard
17:13 sapier first of all the surface detection function needs to be added
17:13 sapier next is automatic animation speed adjustment as well as automatic rotation adjustment
17:14 sapier both will improve player experience drasticaly as they smoothen mob movement on client
17:14 sapier kahrl http://pastebin.com/6MTcpkqs
17:37 kahrl sapier: I was confused earlier because I thought the functions that gave errors were zlib functions
17:38 kahrl but I think I know what's happening now
17:38 kahrl ZLIB_WINAPI is different between where those functions are defined (unzip.c) and where they are used (guiLuaApi.cpp)
17:38 sapier hmm easy to test
17:40 sapier but sounds reasonable
17:53 sapier kahrl the linker errors are gone but it still crashes :-/
17:55 smoke_fumus joined #minetest-dev
17:55 kahrl :| must be something unrelated then
17:56 kahrl do you have a stacktrace?
17:56 sapier #0 0x0334ac73 in ?? ()
17:56 sapier #1 0x00000000 in ?? ()
17:56 kahrl helpful
17:56 sapier so not really a stack :-)
17:57 sapier and this is already a debug build
17:57 sfan5 stack corruption?
17:58 sapier don't have valgrind on win32 to proof or deny
17:58 kahrl where do --verbose output and debug.txt stop?
17:58 sapier Irrlicht engine version 1.7.1
17:59 sapier I don't see any debug.txt ... hmm maybe this isn't a debug build as I wanted to have
18:00 kahrl release builds create one too
18:00 sapier and where?
18:00 Exio ~/.minetest/!
18:01 sapier windows!
18:01 Calinou (run in place)
18:01 Calinou non run in place, maybe in %appdata%
18:04 sapier found at apdata 0 bytes none at run in place
18:05 sapier doesn't anyone have a proven to work win32 environment? :-)
18:05 Calinou no
18:06 sapier funny we support win32 env but there seems to be exactly one environment where build (by accindent) works?
18:27 sapier same result running witing gdb in eclipse ... ok not surprising
18:30 Anchakor joined #minetest-dev
18:51 sapier great ... it's irrlicht 1.7.1 ... once compiling against 1.8 it's working
18:56 dysoco Aw, was hoping I could develop from Windows
18:56 dysoco I have a Linux box, although lately I haven't been booting it
18:57 sapier you can
18:57 sapier do it!
18:57 dysoco I can then ?
18:57 dysoco Or I can boot it?
18:57 sapier we need ppl knowing how to build on windows ;-)
18:57 dysoco Ah
18:57 dysoco developing in Windows is a pain, really
18:58 dysoco but a Minetest Visual C++ project wouldn't be that bad...
18:58 thexyz sapier: everything is written in readme
18:58 sapier I know :-) but if you want to do so you'll have unique knowledge noone else has :-)
18:58 thexyz how's that?
18:59 sapier no it's not there are still some minor things that will make your build fail
19:00 dysoco The info is outdated
19:00 dysoco and last time I tried it failed
19:00 thexyz okay
19:00 sapier that's result of none developing on windows
19:01 sapier e.g. irrlicht 1.7.1 build fails without usable error message or debug stack
20:30 MJaoune joined #minetest-dev
20:30 serengeor joined #minetest-dev
20:35 MJaoune left #minetest-dev
21:03 emptty1 joined #minetest-dev
21:11 mrtux_ joined #minetest-dev
21:40 sapier joined #minetest-dev
21:47 sapier https://github.com/sapier/minetest/tree/ngmm_minizip this works on windows ...
21:47 sapier if you use irrlicht 1.8 and you have to use gettext as well as libarchive from msys
21:48 sapier or disable libarchive support
21:48 PilzAdam my win builds use Irrlicht 1.7.2
21:48 sapier maybe
21:48 sapier I didn't try 1.7.2
21:48 PilzAdam and I dont really want to switch to 1.8
21:48 sapier i can only tell it doesn't run with 1.7.1
21:49 thexyz PilzAdam: that's not the reason to not include it
21:49 sapier try with 1.7.2 maybe it's working too
21:49 PilzAdam well, I though we support Irrlicht 1.7.* and 1.8
21:49 PilzAdam *thought
21:50 thexyz for linux
21:50 sapier minetest needs polishing for 1.8 textures are a little bit disorten
21:50 sapier at least fullscreen textures
21:50 thexyz do you really think one will want us to support <1.8 irrlicht on windows?
21:52 PilzAdam well, fix the buildbot in utli/buildbot/ then
21:52 sapier is there any reason why we don't use gettext and libz from msys?
21:53 thexyz PilzAdam: -irrlicht_version=1.7.2 +irrlicht_version=1.8
21:53 sapier wait
21:53 PilzAdam thexyz, last time I checked I needed to more .dlls
21:53 PilzAdam *two
21:53 sapier no
21:54 sapier it's same
21:54 sapier but built with 1.8 results in graphics errors
21:54 sapier at least in main menu
21:54 PilzAdam thexyz, and there are all the checkboxes and input fields that are somehow glitched out
21:54 sapier scaling seems to be different in 1.8
21:54 thexyz PilzAdam: really?
21:54 thexyz I didn't notice that in official build
21:55 thexyz which uses 1.8
21:55 PilzAdam its in the official build too
21:55 sapier is official build vs or mingw?
21:55 PilzAdam msvc
21:55 thexyz how sad
21:57 thexyz PilzAdam: is this bug reported?
21:57 PilzAdam its known AFAIK
21:57 thexyz PilzAdam: issue #?
21:58 PilzAdam there is no issue yet
21:58 PilzAdam Ill create one
22:00 PilzAdam https://github.com/minetest/minetest/issues/765
22:04 thexyz PilzAdam: http://i.imgur.com/Ee3Soye.png
22:04 thexyz official 0.4.7 in wine
22:04 sapier looks fine ?
22:04 thexyz yes
22:05 thexyz mine does, at least
22:05 thexyz dunno what's wrong with PilzAdam'
22:05 thexyz +s
22:06 sapier it's disorted for me too thexyz
22:06 PilzAdam there is a screenshot at the issue
22:06 MJaoune joined #minetest-dev
22:06 sapier but it's even worse
22:07 thexyz PilzAdam: I see
22:07 sapier and I can't compare to 1.7 as 1.7.2 doesn't work for me either
22:07 thexyz PilzAdam: but it works fine for me in wine and in windows
22:08 thexyz PilzAdam: what's your environment?
22:08 PilzAdam its always like that for me with Irrlicht 1.8 in Linux and wine
22:09 PilzAdam and with 1.7.2 in wine and 1.7.3 in Linux its normal
22:10 thexyz then we should ask others to test it
22:14 thexyz especiall windows users
22:15 sapier I can only test 1.8
22:21 MNJaoune joined #minetest-dev
22:22 Flux1234 joined #minetest-dev
22:23 sapier http://imgur.com/Z2LYQ60
22:23 sapier this is what happens for me with 1.8
22:25 thexyz don't see anything wrong apart from the logo
22:25 sapier yes this happens to all background textures
22:26 PilzAdam selected input field is completly green and checkboxes are grey instead of white
22:26 PilzAdam same issue as 765
22:26 sapier I guess this isn't an error but a missing color setting not beeing present in 1.7?
22:27 thexyz uhhh?
22:27 sapier colors are set by skin
22:27 thexyz wait, so you think the one on the right is the correct?
22:27 PilzAdam of course
22:27 thexyz of course not
22:28 PilzAdam it is how it has always been
22:28 sapier I guess this is an error :-)
22:28 PilzAdam its changed when you migrate to 1.8
22:28 kahrl I prefer them consistent with the other input fields
22:29 thexyz ughh
22:29 thexyz PilzAdam: you're wrong
22:30 thexyz PilzAdam: https://github.com/minetest/minetest/commit/543fc6702f4ae9ad4dbc8289b8c3ed7d8b332cb1
22:31 sapier :-) not a bug but feature
22:31 PilzAdam well, then change that
22:31 thexyz I've never said it was a bug
22:31 thexyz change what?
22:31 thexyz it's how it's supposed to be
22:31 thexyz PilzAdam: your irrlicht is probably pre-1.8
22:31 PilzAdam I used the official win build
22:32 thexyz the official win build is at the left, correct?
22:32 PilzAdam yep
22:32 thexyz it looks just as it's supposed to look
22:32 thexyz see the commit I linked
22:32 PilzAdam yea, but it still looks wrong
22:33 thexyz fuck
22:33 thexyz i'm done with that
22:33 sapier :-) I guess we've got a interpretation problem
22:33 thexyz good night, all
22:34 sapier pilzadam imho the new one is correct as it's consistent it's just a missing feature in 1.7 this can't be fixed there
22:35 PilzAdam how is it "more consistent" in 1.8?
22:35 sapier input fields are same and highlighting is really "highlighted" the way it was before isn't that obvious
22:36 sapier maybe we should add the cursor in new version too I guess it's color is set wrong too
22:37 sapier still I wonder what is wrong with that texture in my build :)
22:40 PilzAdam sapier, I saw that in some youtube videos of previous MT win builds too, IIRC
22:41 sapier strange
22:41 sapier i thought it'd be a result of my mainmenu changes but if this can happen without it too ... hmm
22:46 thexyz http://forum.minetest.net/viewtopic.php?pid=94144#p94144
22:46 thexyz oh wait, I forgot I'd gone to sleep
23:00 celeron55 joined #minetest-dev
23:50 Taoki joined #minetest-dev

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