--- Log opened Sun Apr 05 00:00:33 2015
20150405 00:01:04< iceiceice> i make a lot of patches that i'm not sure if i will merge them, it depends if i think its actually good enough when i'm done
20150405 00:01:14< iceiceice> its fairly painful to revert patches if you decide it was a bad idea
20150405 00:01:48< iceiceice> patch is just an idea :)
20150405 00:11:21< iceiceice> shadowm: thank you for explaining what a trailing space is
20150405 00:14:43-!- trewe [~trewe@2001:8a0:d108:a601:626c:66ff:fe92:9b7c] has quit [Quit: quit]
20150405 00:15:30< vultraz> shadowm: can you yay or nay #386 before 1.13
20150405 00:18:21-!- zookeeper [~lmsnie@wesnoth/developer/zookeeper] has quit [Ping timeout: 265 seconds]
20150405 00:18:24< shadowm> Nay.
20150405 00:19:57< shadowm> iceiceice: So, to recap from last night:
20150405 00:20:08< shadowm> 1) scons seems to require BFS on master but cmake doesn't.
20150405 00:20:39< shadowm> 2) cmake can't compile the dbus notification stuff. This is a blocker for 1.13.0.
20150405 00:21:08< iceiceice> i'm not in charge of maintaining the cmake file
20150405 00:21:09< shadowm> 3) Boost.filesystem/our code using it is wrong in a few regards.
20150405 00:21:24< shadowm> The cmake file is maintained by everyone.
20150405 00:21:31< iceiceice> as far as i'm concerned the cmake file is totally unnecessary
20150405 00:21:38< iceiceice> because no one uses it
20150405 00:21:43< iceiceice> that's also why its always broken
20150405 00:21:45< shadowm> Debian uses it.
20150405 00:21:47< shadowm> I use it.
20150405 00:21:49< iceiceice> debian can use scons
20150405 00:21:50< shadowm> Ivanovic uses it.
20150405 00:21:53< iceiceice> ivanovic is inactive
20150405 00:21:57-!- noy [~Noy@wesnoth/developer/noy] has quit [Quit: noy]
20150405 00:22:02< shadowm> Ivanovic has been using it for all these years.
20150405 00:22:12< iceiceice> i dont think thats true,
20150405 00:22:17< shadowm> Yes, it is true.
20150405 00:22:20< iceiceice> as i recall, cmake is generally broken,
20150405 00:22:24< iceiceice> except once every 3 -6 months
20150405 00:22:31< iceiceice> when ivanovic comes online to build and discovers it doesnt work
20150405 00:22:47< iceiceice> if he wants to build the project, he should use what the active developers use
20150405 00:22:47< shadowm> No, it just started breaking when you started making changes without testing them.
20150405 00:22:51< iceiceice> no, thats' not true
20150405 00:22:55< iceiceice> and its not my job to test with cmake
20150405 00:23:23< shadowm> It's not your job to test with cmake but it's your job to inform us if you aren't sure about some change you are making to it.
20150405 00:23:24< iceiceice> just like its not gfgtdf's job to test with scons or cmake
20150405 00:24:15< iceiceice> shadowm: you should assume that the most that i will generally touch the cmake file,
20150405 00:24:18< iceiceice> is to add new source file entries
20150405 00:24:18< shadowm> scons is a PITA to work with for out-of-tree builds, so I'll do as Ivanovic and keep using cmake for releases.
20150405 00:24:26< iceiceice> its really not
20150405 00:24:32< iceiceice> and if so, then we should fix it
20150405 00:24:58< iceiceice> it is really stupid to use cmake for builds when all the active devs are using scons, the continuous integration uses scons, and nobody uses cmake
20150405 00:25:04< iceiceice> why not just release what the devs are actually building
20150405 00:25:13< shadowm> mordante uses cmake too.
20150405 00:25:15< iceiceice> instead of fighting this pointless political battlew
20150405 00:25:19< iceiceice> mordante is inactive also.
20150405 00:25:25< shadowm> He just happened to go inactive around the same time people started breaking it.
20150405 00:25:49< iceiceice> i didnt see any substantial commits from mordante in more than a year
20150405 00:26:36< iceiceice> look its pretty simple,
20150405 00:26:41< shadowm> Look, if you don't like cmake, fine, but tell us when you are going to move things around and not bother testing it.
20150405 00:26:48< iceiceice> im not interested in fixing the numerous problems with the cmake script
20150405 00:27:00< shadowm> If you can use scons, testing cmake isn't that hard.
20150405 00:27:07-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 00:27:09< iceiceice> no, it really is.
20150405 00:27:25< iceiceice> i think it would be a complete waste of my time
20150405 00:27:43< iceiceice> and frankly i think its a complete waste of yours to try to fix it also
20150405 00:28:07< shadowm> That way it sounds like you are trying to sabotage the cmake recipe.
20150405 00:28:25< iceiceice> i didnt sabotage it, i just didnt take responsiblity to maintain it
20150405 00:28:26< shadowm> "I don't like it, so I'll break it and tell people to not bother fixing it."
20150405 00:28:29< iceiceice> thats bull
20150405 00:28:44< iceiceice> it requires an active maintainer to keep working
20150405 00:28:51< iceiceice> since no one uses it and its not tested, it breaks
20150405 00:28:53< iceiceice> that's not my fault
20150405 00:29:09< iceiceice> you cant say "who touched the project last is responsible to fix it"
20150405 00:29:13< shadowm> Let's have travis use it too then.
20150405 00:29:17< iceiceice> any time there is a commit and it wasn't tested its potentially broken
20150405 00:29:44< iceiceice> sure, knock yourself out,
20150405 00:29:47< shadowm> Yes, I can say that as long as that person is around.
20150405 00:29:51< iceiceice> put it on travis and jenkins,
20150405 00:29:55< iceiceice> but you have to fix it first
20150405 00:29:58< iceiceice> if you want to use it
20150405 00:30:07< shadowm> You don't go to a store and break some vases and then claim it's not your responsibility.
20150405 00:30:24< iceiceice> thats a ridiculous analogy
20150405 00:30:30< iceiceice> if you have a software project thats being actively developed
20150405 00:30:42< iceiceice> and some build system is out of date and no one uses it or takes responsibility to maintain it,
20150405 00:30:43< iceiceice> its going to break
20150405 00:30:49< iceiceice> software requires maintannce
20150405 00:31:01< iceiceice> scons works mainly because loonycyborg has put time and effort into keeping it working
20150405 00:31:15< shadowm> cmake has been used for every single release since around 1.7.x.
20150405 00:31:17< iceiceice> i have tried to keep that going, but my contribution is minimal
20150405 00:31:22< iceiceice> cmake has no equivalent maintainer
20150405 00:31:26< shadowm> That's *years* using cmake.
20150405 00:31:35< iceiceice> that's totally irrelevant
20150405 00:31:40< iceiceice> if no one wants to touch it,
20150405 00:31:46< shadowm> The maintainers are Ivanovic and mordante. You don't go around asking loonycyborg to add files to scons for you, do you?
20150405 00:31:46< iceiceice> and no one does touch it for 2-3 years,
20150405 00:31:49< iceiceice> its going to get hopelessly broken
20150405 00:31:53< iceiceice> well, that explains why its broken.
20150405 00:32:00< iceiceice> since they are both inactive
20150405 00:32:24< shadowm> My question still stands.
20150405 00:32:47< iceiceice> its ridiculous that you try to hold *me* responsible
20150405 00:32:54< iceiceice> as opposed to anyone else who has been making commits
20150405 00:33:10< shadowm> You moved the dbus code around.
20150405 00:33:38< iceiceice> you only do that because i'm a nice guy and i often have taken responsibility to fix broken stuff like travis and some of the build system code,
20150405 00:33:49< iceiceice> and you think that if you put enough pressure on me i'll agree to fix cmake
20150405 00:33:58< iceiceice> you can't force me to do something that i dont want to do, and never agreed to
20150405 00:34:04< shadowm> If someone breaks the add-ons manager code I'll look into it as well because I maintain the add-ons manager code.
20150405 00:34:35< shadowm> The notifications code had no maintainer but then you started fixing and refactoring things in it.
20150405 00:35:16< shadowm> We actually all implicitly agree to maintain the cmake and scons recipes.
20150405 00:35:18< iceiceice> if ivanovic and mordante are the cmake maintainers as you claimed, please ask them to look into whatever issues you have with it
20150405 00:35:53< iceiceice> shadowm: i did not ever agree to maintain the cmake recipe,
20150405 00:35:58< iceiceice> and its obviously not true that we all implicitly agreed to it
20150405 00:36:09< iceiceice> as many people use msvc or xcode exlcusively
20150405 00:36:15< shadowm> Well, I did, and I have been asked to fix things before.
20150405 00:36:26< shadowm> Even without knowing much about cmake at all.
20150405 00:36:40< iceiceice> i'm not willing to do it
20150405 00:37:04< loonycyborg> hmm
20150405 00:38:08< loonycyborg> I didn't see ivanovic and mordante for a long time already
20150405 00:38:33< shadowm> iceiceice: Also, why does the INSTALL file mention a "history" library that's actually called readline here?
20150405 00:38:42< loonycyborg> I can fix it too if it's some trivial issue
20150405 00:39:11< shadowm> loonycyborg: 
20150405 00:39:12< shadowm> /home/shadowm/src/wesnoth/src/desktop/dbus_notification.cpp:23:2: error: #error "The HAVE_LIBDBUS symbol is not defined, you do not have lib dbus available, you should not be trying to compile dbus_notification.cpp"
20150405 00:39:31< shadowm> Also, this reminds me, the cmake recipe on master is inexplicably unable to find fribidi while scons is (and cmake on 1.12).
20150405 00:39:59< loonycyborg> maybe you can link the commit that dealt with dbus stuff?
20150405 00:40:47< shadowm> There are a lot of those, I'm not sure where to even begin.
20150405 00:41:21< loonycyborg> but from this message it seems to me that it doesn't properly pass that define to code
20150405 00:41:49< loonycyborg> maybe some moving around of source files could cause this
20150405 00:41:52< shadowm> Should I look at CMakeLists.txt or the code file itself?
20150405 00:42:03-!- Coffee_irc [~david@ppp121-45-36-236.lns20.adl2.internode.on.net] has joined #wesnoth-dev
20150405 00:44:07-!- noy [~Noy@wesnoth/developer/noy] has quit [Quit: noy]
20150405 00:44:12< shadowm> !commit 31487bec2653f8b2424769bc113e28d5bea9943e @ loonycyborg 
20150405 00:44:12< loonycyborg> in lists
20150405 00:44:13< shikadibot> loonycyborg: Revision 31487bec2653 (Chris Beck) on Mon Jul 14 04:48:02 2014:
20150405 00:44:16< shikadibot> loonycyborg: move notifications into desktop namespace, rename folder also
20150405 00:44:18< shikadibot> loonycyborg:  
20150405 00:44:21< shikadibot> loonycyborg: We also move the desktop_utils files to desktop/open.?pp, and
20150405 00:44:23< shikadibot> loonycyborg: (+1 discarded lines)
20150405 00:44:26< shikadibot> loonycyborg: Web interface URL: https://github.com/wesnoth/wesnoth/commit/31487bec2653
20150405 00:44:54< iceiceice> shadowm: history is a separate library from readline,
20150405 00:45:02< iceiceice> it seems to be packaged with readline in ubuntu
20150405 00:45:04< iceiceice> https://launchpad.net/ubuntu/+source/readline6/6.2-9ubuntu1
20150405 00:45:13-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 00:45:17< shadowm> loonycyborg: Commit d412e89fed8562cbaa9090dd5f216b4b5b86f743 might be relevant too, not sure.
20150405 00:45:22< iceiceice> but its a standalone thing that can be compiled separately, as i did once
20150405 00:45:23< iceiceice> http://linux.die.net/man/3/history
20150405 00:45:33< shadowm> iceiceice: It's libreadline on Debian.
20150405 00:46:16< iceiceice> yeah but they are pretty clear that the package is two libraries
20150405 00:46:17< iceiceice> https://packages.debian.org/wheezy/utils/readline-common
20150405 00:46:24< shadowm> Does any distribution actually make a distinction? I think whatever Debian/Ubuntu use should be given precedence in documentation anyway, since that's what all the cool kids use these days.
20150405 00:46:59< shadowm> The package's description isn't as important as the naming pattern.
20150405 00:47:10< iceiceice> if you want to put notes in install poitning out that its "readline-dev" in debian and ubuntu, that's probably a good idea
20150405 00:47:18< shadowm> i.e. there's no "libhistory" here, but there's the libreadline* packages.
20150405 00:48:24< shadowm> iceiceice: I'm asking again, is BFS mandatory for scons on master?
20150405 00:48:32< iceiceice> yes
20150405 00:48:46< shadowm> I'm asking because: 1) it's not for cmake; 2) INSTALL still says it's optional.
20150405 00:49:06< iceiceice> yes, for 1.12 its optional
20150405 00:49:21< shadowm> Okay, so I guess I have to backport my tests.
20150405 00:50:00< iceiceice> depends if they are C++ or wml unit tests,
20150405 00:50:05< iceiceice> if they are wml unit tests then they cant be backported
20150405 00:50:35< shadowm> C++, bug #23216.
20150405 00:52:56< shadowm> God I hate the Boost.test documentation.
20150405 00:54:23-!- noy [~Noy@wesnoth/developer/noy] has quit [Quit: noy]
20150405 00:54:29< shadowm> Okay, several tests yield different results with the legacy implementation. God dammit.
20150405 00:55:58< loonycyborg> I saw some flamewars about boost.test documentation on boost ml
20150405 00:56:21< loonycyborg> they heckled poor maintainer to death :(
20150405 00:56:56< loonycyborg> iirc they have next generation docs in dev branch :P
20150405 00:57:47< shadowm> I'm not bothered by the author's questionable language skills, but rather about its structure.
20150405 00:58:12< loonycyborg> yes that's exactly what it was about
20150405 00:59:26< shadowm> Sigh: http://pastebin.com/rM0VEp5b
20150405 00:59:31-!- irker696 [~irker@uruz.ai0867.net] has quit [Quit: transmission timeout]
20150405 01:00:12< shadowm> Let's see, 1 is clearly wrong.  `basename /` gives /.
20150405 01:01:44< shadowm> 2 is also wrong.
20150405 01:02:06< shadowm> 3 is a special one because both the legacy implementation and BFS are wrong. `dirname foo/bar/` gives "foo".
20150405 01:02:42< shadowm> 4 is wrong.
20150405 01:03:03< shadowm> 5 is correct, but BFS is wrong.
20150405 01:03:46< shadowm> 6 seems to reveal a bug with the BFS implementation that I assumed was a feature.
20150405 01:04:23< iceiceice> shadowm: here was the email: https://mail.gna.org/public/wesnoth-dev/2014-12/msg00022.html
20150405 01:04:30< shadowm> gfgtdf, iceiceice: Opinions? The paste above is the results with the legacy implementation on 1.12.
20150405 01:04:54< shadowm> Between brackets there's the legacy result vs. the BFS result.
20150405 01:06:17< shadowm> I wouldn't bother fixing the legacy implementation since odds it's as incorrect in 1.12 as it was in 1.10, but cases 3 and 6 reveal problems with the BFS implementation.
20150405 01:06:52< iceiceice> so 1-5 its about having a trailing / ?
20150405 01:07:08< iceiceice> and 6 is about giving a full path when i guess you think it should only be relative?
20150405 01:07:28< shadowm> Right, 3, 5 and 6 are the problematic BFS cases.
20150405 01:07:42< iceiceice> why does there need to be a trailing /
20150405 01:07:52< shadowm> Okay, 3 shows that the BFS implementation of directory_name is incongruous with Unix.
20150405 01:08:01-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 01:08:43< shadowm> BFS gives "foo/bar" where Unix dirname gives foo (both the GNU coreutils and FreeBSD versions of dirname).
20150405 01:09:32< shadowm> 5, the BFS implementation of directory_name gives an empty string, Unix dirname gives /.
20150405 01:09:57< shadowm> Which makes sense by all accounts since / is a directory by itself.
20150405 01:10:04< shadowm> It's also its own parent.
20150405 01:10:56< iceiceice> i dont have a detailed enough knowledge of wesnoth filesystem code to see why having a trailing / or not should be a major bug
20150405 01:11:06< iceiceice> i mean if the game is not failing to load resources, surely its fine
20150405 01:11:10< shadowm> I'm not sure what's going on with 6, but while the previous two cases might sound like unnecessary pedantry since we are not exactly writing a system shell here (though it shows that Boost.filesystem might not be the best tool to write one such), this one might have implications for us now or in the future.
20150405 01:11:42< iceiceice> 6 sounds bad, like potentially a security problem i guess,
20150405 01:11:48< iceiceice> but only if you can actually exploit it
20150405 01:12:02< iceiceice> i seem to remember that gfgtdf tried hard to prevent using .. and similar to escape the wml root
20150405 01:12:55-!- noy [~Noy@wesnoth/developer/noy] has quit [Ping timeout: 256 seconds]
20150405 01:15:25< shadowm> I thought we had fixed that entirely in 1.5.x.
20150405 01:15:37< shadowm> Or 1.7.x, I don't remember which.
20150405 01:16:41-!- gfgtdf_ [~chatzilla@f054143211.adsl.alicedsl.de] has joined #wesnoth-dev
20150405 01:17:39< shadowm> The WML preprocessor uses get_wml_location() but passes a starting directory instead of an empty string.
20150405 01:19:05< shadowm> Since I can't include INSTALL from data/_main.cfg, I presume the problem is with any invocation where the starting directory is an empty string.
20150405 01:19:30-!- gfgtdf [~chatzilla@x55b1aed9.dyn.telefonica.de] has quit [Ping timeout: 252 seconds]
20150405 01:19:36-!- gfgtdf_ is now known as gfgtdf
20150405 01:21:34< shadowm> filesystem::read_map(), the core module loading code in game_config_manager.cpp, load_file() and intf_have_file() in src/scripting/lua_fileops.cpp, some parts of the AI engine.
20150405 01:21:46< shadowm> All these call get_wml_location() with unsanitized output without a starting path.
20150405 01:21:54< shadowm> Er, unsanitized input.
20150405 01:22:35< shadowm> iceiceice: What's load_file() used for?
20150405 01:23:11< iceiceice> you mean in lua?
20150405 01:23:21< iceiceice> i think it implements wesnoth.require and wesnoth.do_file
20150405 01:23:36< iceiceice> in master right?
20150405 01:23:48< shadowm> Yes, I don't think lua_fileops.cpp exists in 1.12.
20150405 01:24:26< shadowm> Whatever it is, it's a potential attack vector.
20150405 01:25:20< shadowm> Hm, what does it mean when executing this `lua print(wesnoth.have_file("./INSTALL"))` I get:
20150405 01:25:23< shadowm> 20150404 22:24:52 error scripting/lua: In function lua_kernel::run(): When executing, Lua runtime error:
20150405 01:25:26< shadowm>   basic_string::_S_construct null not valid
20150405 01:25:45< iceiceice> it sounds pretty wierd
20150405 01:25:54< iceiceice> i think basic_string::_S_construct null is a C++ error
20150405 01:26:06< iceiceice> caused by trying to construct std::string(NULL)
20150405 01:26:24< shadowm> Okay, so it's Lua catching a C++ exception.
20150405 01:26:29< iceiceice> it either means that,
20150405 01:26:37< iceiceice> or it means i messed up some part of the lua error reporting
20150405 01:26:44< iceiceice> by not checking null when lua might give null or something
20150405 01:27:01< iceiceice> i think it might be the former, but i would check a backtrace to be sure if possible
20150405 01:27:01< shadowm> So how do I actually print the return value of that function call? :p
20150405 01:27:20< iceiceice> in lua_kernel::run....
20150405 01:28:06< shadowm> I forgot how to Lua.
20150405 01:28:39< shadowm> Okay, I brought up the Lua console and entered the plain expression instead.
20150405 01:28:49< shadowm> wesnoth.have_file("./INSTALL") yields 'true'.
20150405 01:29:44< shadowm> Is there a direct command to access the Lua console without clicking on the inspector dialog first?
20150405 01:29:55< iceiceice> yes
20150405 01:29:57< iceiceice> its a hotkey
20150405 01:29:59< iceiceice> by default `
20150405 01:30:26< shadowm> When running Wesnoth with $HOME as CWD, "src/wesnoth/wesnoth" yields false but "./src/wesnoth/wesnoth" yields true.
20150405 01:30:27< iceiceice> also you can use "wesnoth.show_lua_console" i think
20150405 01:30:28< iceiceice> or similar
20150405 01:30:48< shadowm> That's /home/shadowm/src/wesnoth/wesnoth, the Wesnoth executable image.
20150405 01:31:42< iceiceice> sounds like... not a good thing :(
20150405 01:32:05< shadowm> I've always wondered why games tend to use ` as a shortcut for some sort of in-game console. It usually doesn't work for me because on Spanish keyboards ` is a dead key, and not even first-level.
20150405 01:32:20< iceiceice> i dont know, i only did it because i saw other games do it :p
20150405 01:32:32< shadowm> So I wind up remapping it to backspace or such.
20150405 01:32:48< iceiceice> i think its just some key thats often useless and gets mapped to consoles
20150405 01:34:00< shadowm> $ wesnoth.dofile("./test.lua")
20150405 01:34:02-!- Appleman1234 [~Appleman1@s108.BMT-e1.vectant.ne.jp] has quit [Remote host closed the connection]
20150405 01:34:07< shadowm>     Hello world from /home/shadowm!
20150405 01:34:16< shadowm> Yep, it runs /home/shadowm/test.lua.
20150405 01:35:21< shadowm> Besides this and wesnoth.require, we don't have any other public API to read files from Lua, right?
20150405 01:36:16< gfgtdf> the lua error is most luikeley an error in teh lua_kernel_base::intf_print function
20150405 01:36:18< shadowm> So this has negligible impact because the target file has to be valid Lua and runs on a single hosts.
20150405 01:36:24< shadowm> host.
20150405 01:36:35< shadowm> So any syntax errors reported will only appear to that host.
20150405 01:36:51< shadowm> Right?
20150405 01:37:01< gfgtdf> shadowm: maybe lua code youd catch teh exception and send teh erro message pover nettwork
20150405 01:37:33< gfgtdf> s/youd/could
20150405 01:37:53< shadowm> Should I start erasing bits from this IRC log now or not.
20150405 01:38:56< shadowm> Before I try to figure out where the BFS version of get_wml_file() went wrong, if BFS is mandatory on master, then why is the legacy version still in src/?
20150405 01:38:58< iceiceice> yeah i think that the only thing thats bad is that you can use havefile
20150405 01:39:07< iceiceice> shadowm: i wrote in my email why
20150405 01:40:04< shadowm> Why is OS X an exception?
20150405 01:40:20< iceiceice> because i cant edit it :p
20150405 01:40:26< iceiceice> and didn't want to bother mattsc at the time
20150405 01:40:33< shadowm> Have you bothered him since then?
20150405 01:40:41< iceiceice> no, i worked on different things
20150405 01:40:51< iceiceice> i think there was still some residual issues about
20150405 01:40:56< shadowm> Maybe you should bother him and/or ancestral now then.
20150405 01:40:57< iceiceice> wanting to support OS X 10.5 and such
20150405 01:41:07< iceiceice> like, partly we didnt make 1.12 BFS mandatory for that reason
20150405 01:41:09< iceiceice> as well as pandora
20150405 01:41:11-!- kex [~kex@78.157.29.160] has joined #wesnoth-dev
20150405 01:41:30< iceiceice> i dont think we plan to support OS X 10.5 in 1.13,
20150405 01:41:40< shadowm> I've been trying to get in touch with ancestral for next Sunday but haven't had any luck so far, so I'm worried that we may have issues with the OS X package.
20150405 01:42:08< iceiceice> i have a script that can make an OS X executable
20150405 01:42:28< iceiceice> but we never figured out how to make application bundles
20150405 01:42:29< shadowm> Does it make a proper bundle as OS X users as accustomed to, though?
20150405 01:42:37< shadowm> *are
20150405 01:43:10< iceiceice> no, but maybe we could solve that if we are desperate
20150405 01:43:17< shadowm> Yeah, that's the thing. I remember that years ago there was an outrage because our new packager started publishing installers instead of bundles.
20150405 01:43:43< shadowm> Apparently installers are considered the work of the devil in Apple land or something.
20150405 01:44:48< iceiceice> we almost figured it out i think
20150405 01:44:54< iceiceice> there is some old code in the scons script that claimed to make bundles
20150405 01:45:01< iceiceice> but i think they were old format bundles or something
20150405 01:46:00< iceiceice> the basic idea was to use the travis' deployment features
20150405 01:46:19< iceiceice> if you change the travis.yml file to this:
20150405 01:46:19< iceiceice> https://github.com/cbeck88/wesnoth/blob/deploy-os-x/.travis.yml
20150405 01:46:32< iceiceice> then when it finishes it craps out a "release"
20150405 01:46:33< iceiceice> https://github.com/cbeck88/wesnoth/releases
20150405 01:46:45< iceiceice> that contains a binary
20150405 01:47:30< shadowm> The legacy get_wml_location() logic says that if current_dir is empty and filename starts with "./" then the result is game_config::path plus the rest.
20150405 01:47:50< shadowm> And the BFS version has no such safeguard.
20150405 01:48:59< shadowm> if (*fpath.begin() == ".") { result /= path(current_dir) / filename; }
20150405 01:49:50< shadowm> Seems simple enough for me to fix without breaking anything else, hopefully.
20150405 01:50:50< shadowm> BFS' path class overloads operator/ to represent path concatenation. Yuck.
20150405 01:51:46< gfgtdf> shadowm: i think that simply 314425ab0e57 and 9f458d1bb29cef431 are not present in boost fs code
20150405 01:52:15< gfgtdf> shadowm: becasue those function were written before those commits
20150405 01:52:36< shadowm> Okay, that explains why I had a feeling of déjà vu about this.
20150405 01:53:18< shadowm> So that didn't break anything, I don't think I'll break anything this time either.
20150405 01:55:30< gfgtdf> shadowm: but i wonder why we dont so substr(2) in the bf boost code
20150405 01:55:37< gfgtdf> s/so/do
20150405 01:56:19< shadowm> Well, foo/./bar is equivalent to foo/bar, so that isn't an essential step.
20150405 01:56:46< gfgtdf> shadowm: ok didn't know that
20150405 01:57:33< shadowm> Not doing so does make the path uglier, though.
20150405 01:57:46< gfgtdf> shadowm: but the bfs also onyl checks for '.' not for './' wouldnt taht break files that begin with '.'
20150405 01:58:25< shadowm> } else if (*fpath.begin() == ".") {
20150405 01:58:29-!- tomreyn [~tomreyn@megaglest/team/tomreyn] has quit [Quit: tomreyn]
20150405 01:58:45< shadowm> Depends on exactly what that iterator represents. I assume it's a whole path component?
20150405 01:59:31< gfgtdf> ah yes i forgot that.
20150405 02:01:15< shadowm> So the equivalent to the original substr() would be to append (filename.begin(), filename.end()) instead of [filename.begin(), filename.end()) to result, I believe.
20150405 02:01:48< gfgtdf> shadowm: filename is teh strign object i think
20150405 02:01:51< gfgtdf> string
20150405 02:02:00< shadowm> s/filename/fpath/g
20150405 02:03:00< shadowm> Though the statement is `result /= path(current_dir) / filename` in master.
20150405 02:03:09< gfgtdf> shadowm: hm yes, but i actualy dont know how to append a range.
20150405 02:03:59< shadowm> filename.substr(2) would do it but then we break the implicit assumption that path separators might be more than 1 char long, which I presume is part of BFS' design? I don't know.
20150405 02:04:57< shadowm> Meh, who cares about BFS' design, this function already uses filename.substr() in a previous block.
20150405 02:05:08< shadowm>     result /= get_user_data_path() / "data" / filename.substr(1);
20150405 02:05:29< gfgtdf> shadowm: but that not for a path seperator
20150405 02:05:45< gfgtdf> shadowm: actualy i dont know a nfs that uses multi hex seperators
20150405 02:05:46< shadowm> Eh, yeah, you are right.
20150405 02:06:12< gfgtdf> shadowm: and since that sounds quet unhandy i doulbt there is one.
20150405 02:07:20< shadowm> Actually, I'll just leave it as is, no substr.
20150405 02:07:45< shadowm> Also, I'll have to check what the original code does when fed "." alone.
20150405 02:08:37< shadowm> Wait what, I already know what it does.
20150405 02:08:57< shadowm> It gives "/home/shadowm/src/wesnoth-1.12/data/.".
20150405 02:09:37< shadowm> Now it gives "/home/shadowm/src/wesnoth/.". Still incongruous.
20150405 02:11:08< shadowm> It seems the BFS version of this case was plain broken.
20150405 02:12:13< shadowm> But where's the non-BFS version getting the data/ component from? :\
20150405 02:12:45-!- Implementor37 [44b792a3@gateway/web/freenode/ip.68.183.146.163] has joined #wesnoth-dev
20150405 02:13:07< shadowm> Okay, it's the fallback case on filesystem.cpp:1037, right.
20150405 02:13:37< shadowm> But that means that for the non-BFS version, "_main.cfg" and "./_main.cfg" refer to different locations.
20150405 02:14:00< shadowm> "<datadirroot>/data/_main.cfg" and "<datadirroot>/_main.cfg", respectively.
20150405 02:14:43< shadowm> So I have to replicate this absurd behavior? Argh.
20150405 02:16:18< gfgtdf> shadowm: you cna make curent_dirs default to "<datadirroot>/data" instead of "<datadirroot>" but i dont really care
20150405 02:17:30< gfgtdf> shadowm: what do you think is teh expexted behviour on passing "." ?
20150405 02:17:31< shadowm> Well, let's put it like this. This only affects specific invocations of Lua have_file, dofile, and require that currently break the filesystem glass cage.
20150405 02:17:46< gfgtdf> shadowm: the same as '.\' ?
20150405 02:17:52< shadowm> People won't be using those three with mainline files.
20150405 02:18:22< shadowm> Internally, the engine may call get_wml_location() without an starting dir to load files like hardwired/language.cfg and that's it.
20150405 02:18:42< shadowm> gfgtdf: That's how it should be in a sane world, yes.
20150405 02:19:15< shadowm> I presume this bizarre difference in semantics in the legacy implementation is just a bug no-one has ever encountered.
20150405 02:19:33< shadowm> Until I decided to write tests for the fs API.
20150405 02:20:48< gfgtdf> shadowm: we coudl also try to get teh path of teh currently executed lua file and pass that files directory to get_wml_location.
20150405 02:20:48< shadowm> So I've decided, <dataroot>/data/ it is.
20150405 02:21:43< shadowm> gfgtdf: That would be a sensible thing to do (when not running those functions from the console anyway), but I think iceiceice knows better how the Lua stuff works atm than anyone else.
20150405 02:21:43-!- iceiceice [~chris@wesnoth/developer/iceiceice] has quit [Read error: Connection reset by peer]
20150405 02:22:09-!- iceiceice [~chris@wesnoth/developer/iceiceice] has joined #wesnoth-dev
20150405 02:23:25< shadowm> I think I also found a preprocessor read-past-end-of-buffer bug.
20150405 02:24:02< shadowm> At least with a #warning with an accompanying message that includes EOF.
20150405 02:25:22< gfgtdf> iceiceice: can you fix the error in teh lua print function ?
20150405 02:28:17< gfgtdf> i sleep now
20150405 02:28:19-!- gfgtdf [~chatzilla@f054143211.adsl.alicedsl.de] has quit [Quit: ChatZilla 0.9.91.1 [Firefox 36.0.4/20150320202338]]
20150405 02:30:17< shadowm> I forgot to assess the implications for the AI engine.
20150405 02:33:21< shadowm> Hm, crap.
20150405 02:33:40< shadowm> Anyone feeling like releasing 1.12.2 in a week?
20150405 02:35:17< shadowm> iceiceice: It seems you committed an important fix since 1.12.1 to 1.12.
20150405 02:35:54< iceiceice> i dont remember
20150405 02:36:04< iceiceice> main thing i remember is, should figure out about this server thing
20150405 02:36:05< shadowm> 1627737166e0b52258d7110befab4ad4134de4f2
20150405 02:36:26< iceiceice> should figure out what to do about this:
20150405 02:36:27< iceiceice> https://github.com/wesnoth/wesnoth/pull/379
20150405 02:36:40< shadowm> I need mattsc or someone else who knows the AI code to give me a hand.
20150405 02:37:04< iceiceice> yeah that was a good fix
20150405 02:37:09< iceiceice> i think i changelogged it
20150405 02:39:29< iceiceice> loonycyborg: i get errors in scons , i think maybe the most recent change is not unicode friendly?
20150405 02:39:30< iceiceice> http://pastebin.com/GQkS5pPS
20150405 02:39:33< iceiceice> (i'm not sure how important this is)
20150405 02:39:42< shadowm> Sent a PM to mattsc asking for help.
20150405 02:40:09< shadowm> iceiceice: So you feel we should wait a bit for 1.12.2?
20150405 02:40:09< iceiceice> i can build at HEAD^^
20150405 02:40:34< iceiceice> (i only have wesnoth in wierd path to help catch the filesystem unicode errors from before)
20150405 02:41:05< loonycyborg> iceiceice: hmm didn't expect python would have problems with this..
20150405 02:44:06< iceiceice> shadowm: idk, i will try to fix the server thing i guess and test it
20150405 02:44:17< iceiceice> i think its not essential for release but it would be nice
20150405 02:44:26< shadowm> Okay.
20150405 02:44:28< iceiceice> it might be server only patch anyways
20150405 02:44:32-!- irker477 [~irker@uruz.ai0867.net] has joined #wesnoth-dev
20150405 02:44:32< irker477> wesnoth: Ignacio R. Morelle wesnoth:master 1c927c8407ec / src/filesystem_boost.cpp: fs: Prevent the BFS version of get_wml_location() from escaping the data dir http://git.io/ven54
20150405 02:44:33< irker477> wesnoth: gfgtdf wesnoth:23216 afab09aae934 / RELEASE_NOTES: Update RELEASE_NOTES http://git.io/veYpI
20150405 02:44:34< irker477> wesnoth: Elvish_Hunter wesnoth:23216 157066c51241 / changelog src/unit.cpp: Fixed not being able to set unit.ellipse variable if ellipse= is set by [unit_ty http://git.io/vesJZ
20150405 02:44:35< irker477> wesnoth: loonycyborg wesnoth:23216 3c7a3e648488 / scons/sdl.py: scons: fix config checks that require running test programs for out-of-tree buil http://git.io/veGJx
20150405 02:44:36< irker477> wesnoth: gfgtdf wesnoth:23216 df11117ba47b / src/menu_events.cpp: make :n and :cl end the scenario in linger mode http://git.io/veGsE
20150405 02:44:37< irker477> wesnoth: Ignacio R. Morelle wesnoth:23216 1c927c8407ec / src/filesystem_boost.cpp: fs: Prevent the BFS version of get_wml_location() from escaping the data dir http://git.io/ven54
20150405 02:44:40< irker477> wesnoth: Ignacio R. Morelle wesnoth:23216 fee306087958 / / (6 files in 3 dirs): Merge branch 'master' into feature/23216 http://git.io/ven5B
20150405 02:44:43< irker477> wesnoth: Ignacio R. Morelle wesnoth:23216 57c89ccecb87 / src/tests/test_filesystem.cpp: [DO NOT MERGE] Update tests after bug fix http://git.io/ven5R
20150405 02:44:46< irker477> wesnoth: Ignacio R. Morelle wesnoth:1.12 001d20e2ff5e / src/filesystem_boost.cpp: fs: Prevent the BFS version of get_wml_location() from escaping the data dir http://git.io/ven50
20150405 02:47:37< loonycyborg> shadowm: uhm, I couldn't reproduce your cmake issue
20150405 02:47:51-!- Necrosporus_ [~Necrospor@unaffiliated/necrosporus] has joined #wesnoth-dev
20150405 02:47:59< loonycyborg> did a build in /tmp and it worked
20150405 02:48:09< shadowm> master?
20150405 02:48:11-!- mattsc [~mattsc@wesnoth/developer/mattsc] has joined #wesnoth-dev
20150405 02:48:22< loonycyborg> found dbus-1 with pkg-config
20150405 02:48:35< shadowm> Did you try to compile?
20150405 02:48:48< loonycyborg> yes
20150405 02:48:53< loonycyborg> it ran to completion
20150405 02:49:11< loonycyborg> and yes
20150405 02:49:14< loonycyborg> I'm on master
20150405 02:49:18< loonycyborg> just double checked
20150405 02:49:30< mattsc> shadowm: hi - I don’t have time to read up on the logs to see what this is about right now, but since you said ‘urgent’ …
20150405 02:49:37< shadowm> loonycyborg: Try ENABLE_TESTS.
20150405 02:49:40< loonycyborg> it passed -DHAVE_LIBDUS too
20150405 02:49:53< loonycyborg> ok
20150405 02:50:04< shadowm> Failing that, try ENABLE_GAME=OFF.
20150405 02:50:22< mattsc> If the questions is: is it possible to read something specified in WML (a variable?) with the Lua AI and sync the content to other hosts, then the answer is yes.
20150405 02:50:33< shadowm> loonycyborg: Also, does it find fribidi for you?
20150405 02:51:05-!- Necrosporus [~Necrospor@unaffiliated/necrosporus] has quit [Ping timeout: 265 seconds]
20150405 02:51:10< iceiceice> shadowm: if ENABLE_GAME=OFF then you should not be trying to build dbus notifications i think
20150405 02:51:17< iceiceice> since that is only used for the game
20150405 02:51:26< mattsc> You can do pretty much anything you want with ai.synced_command(); http://wiki.wesnoth.org/Customizing_AI_in_Wesnoth_1.8#Lua_AI_syntax
20150405 02:51:27< shadowm> mattsc: The question is whether it's possible to read it directly from the filesystem, and I'd prefer to see the answer in query.
20150405 02:51:29< iceiceice> so maybe it means that dbus_notifications is listed in wesnoth_core when it should be in wesnoth_lib or something
20150405 02:51:40< iceiceice> for cmake i mean
20150405 02:51:52< loonycyborg> no it doesn't find fribidi
20150405 02:52:01< shadowm> ENABLE_GAME=OFF works for me as long as I don't ENABLE_TESTS=ON.
20150405 02:52:33< shadowm> I'm not sure how enabling tests pulls that file.
20150405 02:53:42< mattsc> shadowm: I’m apparently having troubling parsing sentence tonight.  I don’t know what “I'd prefer to see the answer in query” means.
20150405 02:53:54< loonycyborg> irc pm
20150405 02:53:59< loonycyborg> /query
20150405 02:54:03< shadowm> What loonycyborg said.
20150405 02:54:07< mattsc> oh
20150405 02:54:10< mattsc> duh
20150405 02:54:46< shadowm> Back in 45 minutes.
20150405 02:55:14< shadowm> mattsc: Also, do you know much about the formula AI engine?
20150405 02:55:36< shadowm> This is totally not related to the same question. (It actually is.)
20150405 03:10:53< loonycyborg> shadowm: uhh, test depend on game
20150405 03:11:05< loonycyborg> it doesn't make sense to compile them with game disabled
20150405 03:13:53< loonycyborg> at least they definitely depend on most of object files that are part of game
20150405 03:13:56-!- bumbadadabum [~bumbadada@d155109.upc-d.chello.nl] has quit [Ping timeout: 250 seconds]
20150405 03:18:08< shadowm> Huh.
20150405 03:18:46< shadowm> Well, can you make it so the cmake recipe bails out during configuration or whatever then?
20150405 03:19:37< shadowm> Though I still don't understand how this in particular can be a consequence of disabling the game target.
20150405 03:20:12< shadowm> Does that somehow preclude passing the correct options to the compiler?
20150405 03:21:36-!- c74d [~c74d3a4eb@2002:4404:712c:0:76de:2bff:fed4:2766] has quit [Read error: Connection reset by peer]
20150405 03:23:21< shadowm> And then it still tries to compile object files that are part of the game target? :\
20150405 03:24:44-!- c74d [~c74d3a4eb@2002:4404:712c:0:76de:2bff:fed4:2766] has joined #wesnoth-dev
20150405 03:27:53< loonycyborg> -DENABLE_TESTS=1 -DENABLE_GAME=off
20150405 03:27:56< loonycyborg> still works
20150405 03:28:46< shadowm> Even on a clean dir?
20150405 03:28:52< loonycyborg> yes
20150405 03:29:12< loonycyborg> it doesn't pass -DHAVE_LIBDBUS on command line
20150405 03:29:15< loonycyborg> but succeeds
20150405 03:29:32< shadowm> Do you have dbus installed?
20150405 03:29:37< loonycyborg> yes
20150405 03:29:49< shadowm> And -DENABLE_NOTIFICATIONS=ON.
20150405 03:29:53< loonycyborg> and it is used for all other combinations of options I tried
20150405 03:30:08< shadowm> :|
20150405 03:30:32< loonycyborg> don't have time to try another combination
20150405 03:30:34< loonycyborg> need to sleep
20150405 03:30:35< loonycyborg> nn
20150405 03:31:22-!- iceiceice [~chris@wesnoth/developer/iceiceice] has quit [Quit: Leaving]
20150405 03:43:32-!- Kwandulin [~Miranda@p5B00890E.dip0.t-ipconnect.de] has joined #wesnoth-dev
20150405 03:44:00-!- ancestral [~ancestral@174-20-187-184.mpls.qwest.net] has joined #wesnoth-dev
20150405 03:50:46< shadowm> ancestral: Hi.
20150405 03:51:16< ancestral> shadowm: ohai
20150405 03:51:42< shadowm> So please make sure to have that test build done tomorrow, and don't hesitate to PM mattsc if you need help.
20150405 03:53:57< ancestral> Yeah, mattsc is my go-to for help
20150405 04:08:39-!- mattsc [~mattsc@wesnoth/developer/mattsc] has quit [Quit: Computer's napping]
20150405 04:13:32-!- oldlaptop [~quassel@50-108-82-79.adr01.mskg.mi.frontiernet.net] has quit [Ping timeout: 256 seconds]
20150405 04:17:10-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 04:20:19-!- oldlaptop [~quassel@50-108-82-79.adr01.mskg.mi.frontiernet.net] has joined #wesnoth-dev
20150405 04:21:41-!- iceiceice [~chris@wesnoth/developer/iceiceice] has joined #wesnoth-dev
20150405 04:22:40< iceiceice> shadowm: given the results of filesystem tests,
20150405 04:22:45< iceiceice> does it explain this? http://forums.wesnoth.org/viewtopic.php?f=15&t=22159&start=390#p582897
20150405 04:24:05< vultraz> those relative type paths have never caused me problem (windows)
20150405 04:24:37< vultraz> s/problem/[]s
20150405 04:25:41< shadowm> iceiceice: I don't think so, but I'm not sure.
20150405 04:26:13< iceiceice> yeah i wonder if there's something wierd about doofus-1's setup
20150405 04:27:59< shadowm> Gah.
20150405 04:28:21< shadowm> Okay, I've found something a bit concerning.
20150405 04:31:47< shadowm> Yup.
20150405 04:32:15< shadowm> The bug I just fixed is more serious than I thought.
20150405 04:32:35< vultraz> do tell
20150405 04:33:09< shadowm> http://i.imgur.com/ErmwjLY.png
20150405 04:33:51< shadowm> iceiceice: It's Galactic Empires for 1.12, right?
20150405 04:34:00< iceiceice> i think so
20150405 04:34:21-!- ancestral [~ancestral@174-20-187-184.mpls.qwest.net] has quit [Quit: i go nstuf kthxbai]
20150405 04:34:27< shadowm> Bleh 19.6 MiB.
20150405 04:35:04< shadowm> I'm going to cheat and publish a WML-only tarball for myself.
20150405 04:35:54< iceiceice> oh i forgot to push a patch
20150405 04:36:58< shadowm> It's only 101 KiB without images/music/sounds.
20150405 04:37:09< shadowm> xz'd.
20150405 04:37:22< iceiceice> there's also a ton of repetitive code iirc
20150405 04:37:31< iceiceice> it probably compresses very nicely
20150405 04:37:48< vultraz> that's to be expected
20150405 04:37:50< vultraz> :P
20150405 04:38:04< iceiceice> i think, more repetitive than usual wml 
20150405 04:38:06< shadowm> It's one of those silly add-ons that brace-substitute resource dirs like images/.
20150405 04:38:19 * vultraz groans
20150405 04:38:25< shadowm> And sounds and music.
20150405 04:38:39 * vultraz groans double and heads to wiki
20150405 04:39:07< shadowm> iceiceice: Seems to work for me with a fix-less build.
20150405 04:39:25< iceiceice> i was going to test it, and i got sidetracked by stuff in my working tree
20150405 04:39:35< iceiceice> i tested it fairly recently thoguh
20150405 04:39:41< iceiceice> i wonder if doofus-1 has a strange set-up
20150405 04:40:23< shadowm> It might well be some other add-on polluting things.
20150405 04:40:49< shadowm> Anyway, I've decided we need 1.12.2 ASAP.
20150405 04:40:52< iceiceice> can addons change the current working dir?
20150405 04:41:17< shadowm> Wesnoth's process CWD? No. The preprocessor's? Yes.
20150405 04:41:35< shadowm> The preprocessor's directory is scoped along with everything else, though.
20150405 04:41:46< iceiceice> so how would it happen?
20150405 04:41:53< iceiceice> like, an add-on would include many files into some directory,
20150405 04:41:59< iceiceice> then, try to include {~} or something?
20150405 04:42:07< iceiceice> and then get all the other addons at some wierd part of hte hierarchy?
20150405 04:42:13-!- Implementor37 [44b792a3@gateway/web/freenode/ip.68.183.146.163] has quit [Ping timeout: 246 seconds]
20150405 04:42:29< shadowm> I don't even know what the error doofus-01 encountered is. For all I know it might be some add-on including GE stuff incorrectly.
20150405 04:42:48< shadowm> I don't think {~} is a valid request.
20150405 04:43:13-!- c74d [~c74d3a4eb@2002:4404:712c:0:76de:2bff:fed4:2766] has quit [Ping timeout: 265 seconds]
20150405 04:43:23< iceiceice> yeah i see what you are saying
20150405 04:43:25< shadowm> Hm, okay, it's valid but I don't know what it does.
20150405 04:43:54< iceiceice> probably tries to include your entire data dir?
20150405 04:44:03< iceiceice> i mean thats what i would expect i gues
20150405 04:46:19< shadowm> I'm not getting any output with --log-debug=preprocessor, that's weird.
20150405 04:46:24-!- c74d [~c74d3a4eb@2002:4404:712c:0:76de:2bff:fed4:2766] has joined #wesnoth-dev
20150405 04:46:50< shadowm> Oh, the build I'm using is two days too old for that. :p
20150405 04:48:38< shadowm> 01/10, I moved the relevant output to that domain on 01/12.
20150405 04:49:43< shadowm> 20150405 01:49:35 info preprocessor: Macro definition not found for ~ , attempting to open as file.
20150405 04:50:39< shadowm> That succeeds.
20150405 04:51:09< shadowm> That means get_wml_location("", some_dir) doesn't return empty.
20150405 04:52:06< shadowm> That can't be right.
20150405 04:52:34< shadowm> 20150405 01:52:16 info filesystem: searching for _main.cfg in directory /home/shadowm/.wesnoth-1.13/data
20150405 04:53:00< shadowm> So it doesn't (NORMALLY) find anything and bails out.
20150405 04:54:13< shadowm> From a cursory glance, the legacy implementation will do the same thing only with a trailing slash.
20150405 04:54:55< shadowm> Ugh. Okay, travis hasn't complained about my last test batch, so I'll just merge and backport so I can stop juggling multiple branches and builds.
20150405 04:54:59< vultraz> ah, I already have a note that binary content need not be included
20150405 04:55:38< vultraz> isn't it possible to stick a warning for inclusion of binary directories
20150405 04:57:02< shadowm> The game can't know what directories are "binary directories", though. At most you can guess the contents from the name (one of "images", "sounds", or "music"), but nothing prevents some lunatic from actually sticking a WML file somewhere in there.
20150405 04:57:38< vultraz> so what happens if you preprocess those directories
20150405 04:57:50< shadowm> Nothing at all. The game only looks for *.cfg files.
20150405 04:58:25< vultraz> ah
20150405 04:58:41< irker477> wesnoth: Chris Beck wesnoth:master 91440ba4f385 / src/scripting/lua_kernel_base.hpp: fix 'basic_string::_S_construct' error from lua interpreter http://git.io/vecf9
20150405 04:58:42< shadowm> You just ask the preprocessor to waste a thousand nanoseconds or so on generating a new context, requesting directory entries, matching them against that pattern, coming up with nothing, and destroying the context.
20150405 04:58:46< iceiceice> gfgtdf: ^
20150405 05:03:34 * vultraz wonders why his wesnoth exe alternately outputs 20 Mb and 300 MB
20150405 05:03:55< shadowm> Release vs. debug, right?
20150405 05:04:06< vultraz> No
20150405 05:04:18< vultraz> I always build debug
20150405 05:04:24< irker477> wesnoth: Ignacio R. Morelle wesnoth:1.12 86775d48ee45 / / (14 files in 6 dirs): Add filesystem code unit tests (bug #23216) http://git.io/vecJc
20150405 05:04:27< irker477> wesnoth: Ignacio R. Morelle wesnoth:master 7f5152d0c344 / / (14 files in 6 dirs): Add filesystem code unit tests (bug #23216) http://git.io/vecJC
20150405 05:04:43< shadowm> There's no way a Wesnoth debug build can be 20 MiB unless it's got symbols stripped.
20150405 05:06:08< shadowm> Okay, so what was I going to do next.
20150405 05:09:14< irker477> wesnoth: Ignacio R. Morelle wesnoth:master 464c638b129a / src/tests/test_filesystem.cpp: Remove redundant include http://git.io/vecJ9
20150405 05:11:56< shadowm> So, do we want that weird get_wml_location() behavior with "~"?
20150405 05:12:16< iceiceice> imo, probably no
20150405 05:12:25< shadowm> It doesn't really cause any problems since that path doesn't usually contain any files.
20150405 05:13:02< iceiceice> it always seemed a bit wierd to me that the root path for addons is below data
20150405 05:13:18< iceiceice> so that you have to write {data/addons/my-addon/...}
20150405 05:13:37< iceiceice> i mean i can understand that you might want to include other add-ons, so maybe the root should be at {addons}
20150405 05:13:39< shadowm> I suspect at some point people expected to have more directories than just campaigns/ on <userdata>/data/.
20150405 05:13:45< iceiceice> but we cant change that any ways without breaking compatibility badly
20150405 05:14:19< shadowm> A bit of a historical note: add-ons were stored in <userdata>/data/campaigns/ for three stable series.
20150405 05:14:47< shadowm> It seems like a bit of a logical conclusion to have eras/, maps/, and so on.
20150405 05:15:20< shadowm> But then I came along and renamed it "add-ons" because I decided the alternative would be deemed confusing by the UMC authoring crowd.
20150405 05:16:06< shadowm> On the same note, brace substitutions used to accept @ as a prefix as an alternative to ~.
20150405 05:17:00< shadowm> {@campaigns/Epic_Fail} would include both {campaigns/Epic_Fail} and {~campaigns/Epic_Fail} (IIRC it wasn't a XOR-style choice, but rather both directories were included if they existed).
20150405 05:18:13< shadowm> I vaguely recall a questionably legitimate use for that, but since I renamed ~data/campaigns to ~data/add-ons, it became a moot point and the feature died off.
20150405 05:18:19< shadowm> *there being a
20150405 05:18:46< shadowm> I'm sure vultraz didn't know any of this.
20150405 05:20:16< shadowm> Also, I can't count. It's four stable series.
20150405 05:21:13< shadowm> Anyway.
20150405 05:21:32< shadowm> iceiceice: No, you don't have to write {data/add-ons}. All preprocessor paths are relative to data/.
20150405 05:21:46< iceiceice> hmmm ok
20150405 05:21:59< shadowm> It's {~add-ons/Foo}, {core}, {gui}, and so on.
20150405 05:22:17< iceiceice> so what is this doing? 
20150405 05:22:17< iceiceice> http://forums.wesnoth.org/viewtopic.php?f=15&t=22159&start=390#p582897
20150405 05:22:22< shadowm> If anything, it's an annoyance for people doing work on the command line.
20150405 05:22:25< iceiceice> "path=data/add-ons/Galactic_Empires"
20150405 05:22:39< shadowm> That's not the preprocessor's work.
20150405 05:22:48< iceiceice> hmm ok
20150405 05:23:28-!- noy [~Noy@wesnoth/developer/noy] has quit [Quit: noy]
20150405 05:23:48< shadowm> I think [binary_path]'s value is passed directly to the underlying FS API. Plus one could argue that we have an implicit "root" binary path pointing to the parent of <datadir>/data/.
20150405 05:23:56-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 05:24:15< shadowm> i.e. <datadir>/, home to the top-level fonts/, images/, sounds/, etc.
20150405 05:25:15< shadowm> 	BOOST_CHECK( !is_gzip_file("foo.GZ") );
20150405 05:25:16< shadowm> 	BOOST_CHECK( !is_bzip2_file("foo.BZ2") );
20150405 05:25:35< shadowm> These functions aren't case-insensitive, but looks_like_pbl() is.
20150405 05:26:31< shadowm> I imagine there's little harm done in making them c-i.
20150405 05:27:55< vultraz> shadowm: I believe I did know that
20150405 05:34:07< shadowm> preprocessor_data::read_rest_of_line() stops on EOF, hm.
20150405 05:35:40< shadowm> Right?
20150405 05:36:10< shadowm> Ah, streamed I/O, my one weakness. :/
20150405 05:38:29< shadowm> It *seems* like eof() returns false until you actually try to read the next character?
20150405 05:40:20< shadowm> So it should call peek() first and eof() second.
20150405 05:41:23< shadowm> That's only used by the code handling #error and #warning, and I don't seem to have broken anything and got rid of this weird garbage-at-EOF bug, so it's good, I guess.
20150405 05:44:46< irker477> wesnoth: Ignacio R. Morelle wesnoth:master 8984cefe1d74 / src/serialization/preprocessor.cpp: preproc: Fix garbage at the end of #warning/#error lines at EOF http://git.io/vecku
20150405 05:44:49< irker477> wesnoth: Ignacio R. Morelle wesnoth:master 7c7c94bad94e / src/tests/test_filesystem.cpp: Additional filesystem::get_wml_location() test cases http://git.io/veckz
20150405 05:45:49< irker477> wesnoth: Ignacio R. Morelle wesnoth:1.12 695026c7e620 / src/tests/test_filesystem.cpp: Remove redundant include http://git.io/veck2
20150405 05:45:52< irker477> wesnoth: Ignacio R. Morelle wesnoth:1.12 db0a663858c8 / src/serialization/preprocessor.cpp: preproc: Fix garbage at the end of #warning/#error lines at EOF http://git.io/vecka
20150405 05:45:55< irker477> wesnoth: Ignacio R. Morelle wesnoth:1.12 b46027acde02 / src/tests/test_filesystem.cpp: Additional filesystem::get_wml_location() test cases http://git.io/veckV
20150405 06:08:06< shadowm> iceiceice: I think you were interested in an actual exploit, so you might want to check that.
--- Log opened Sun Apr 05 06:20:27 2015
20150405 06:27:07< shadowm> What.
20150405 06:27:30< shadowm> Oh, never mind, just a reordered string.
20150405 06:31:33-!- shadowm changed the topic of #wesnoth-dev to: String-freeze active for 1.12 branch | Development version 1.13.0 scheduled for April 12th | Wesnoth Developers Channel | Logs: http://irclogs.wesnoth.org
20150405 06:36:01< irker477> wesnoth: Ignacio R. Morelle wesnoth:1.12 e8869e94518b / changelog: Fix the changelog http://git.io/vecmP
20150405 06:36:04< irker477> wesnoth: Ignacio R. Morelle wesnoth:1.12 57da34e797a2 / / (681 files in 28 dirs): pot and documentation update http://git.io/vecmX
20150405 06:41:56< vultraz> I really need to get that windows implemention of git new_working_dir
20150405 06:42:02< vultraz> so I can build 1.12 as well
20150405 06:42:11< shadowm> I've sent an email to the ML concerning 1.12.2.
20150405 06:44:13< vultraz> Interestingly, I was just thinking today of suggesting that
20150405 06:52:36< shadowm> Why are there translatable TODOs in the editor help still argh.
20150405 06:52:54< shadowm> It doesn't really matter, I guess.
20150405 06:53:56-!- Coffee_irc [~david@ppp121-45-36-236.lns20.adl2.internode.on.net] has quit [Quit: Konversation terminated!]
20150405 06:55:02-!- shadowm changed the topic of #wesnoth-dev to: String-freeze active for 1.12 branch | Stable version 1.12.2 scheduled for April 8th/12th | Development version 1.13.0 scheduled for April 12th | Wesnoth Developers Channel | Logs: http://irclogs.wesnoth.org
20150405 07:00:59< irker477> wesnoth: Ignacio R. Morelle wesnoth:master 549a624bfff2 / / (1495 files in 36 dirs): pot and documentation update http://git.io/vecss
20150405 07:02:39< vultraz> are you gonna kill me if I commit strings to master now
20150405 07:02:51< shadowm> It's not string-frozen and won't be until beta 1.
20150405 07:04:31< shadowm> Oh fuck me.
20150405 07:05:51-!- Crendgrim [~crend@wesnoth/forum-moderator/crendgrim] has quit [Quit: No Ping reply in 180 seconds.]
20150405 07:06:09< shadowm> I should've read that email instead of going just by what I asked Ivanovic the other day.
20150405 07:06:17-!- Crendgrim [~crend@wesnoth/forum-moderator/crendgrim] has joined #wesnoth-dev
20150405 07:06:55< shadowm> I have to copy over the translations from 1.12 to master and run another pot-update.
20150405 07:07:33< Ivanovic> iceiceice: i guess one of the issues with the build system(s) is that if you do a change of which you are not 100% if it will work for all build systems, just send an email to wesnoth-dev to ask someone to check to make sure it avtually works
20150405 07:07:56< Ivanovic> since yeah, it is easier to follow this low volume ML to stay up to date regarding what needs to be fixed / adjusted
20150405 07:09:27< iceiceice> in the past i mostly got it right i think,
20150405 07:09:34< iceiceice> the last time missed some punctuation or something
20150405 07:09:52< iceiceice> its not clear that this time anything was actually broken either
20150405 07:10:21< iceiceice> but in general if someone is running the cmake build regularly, its going to break
20150405 07:10:25< Ivanovic> and btw i try to build wesnoth about every other week these days
20150405 07:10:26< iceiceice> *if someone is not
20150405 07:10:37< Ivanovic> both on 64bit as well as 32bit
20150405 07:10:50< iceiceice> news to me
20150405 07:12:28< shadowm> iceiceice: Yeah, sorry, I got frustrated after repeatedly failing to get the builds I wanted on both branches and spending more time fiddling with the build systems instead of coding and testing.
20150405 07:13:05-!- Coffee_irc [~david@ppp121-45-36-236.lns20.adl2.internode.on.net] has joined #wesnoth-dev
20150405 07:13:08< iceiceice> yeah thats understandable
20150405 07:13:41< Ivanovic> and as far as I am aware cmake is also used to create the gentoo builds
20150405 07:13:53< Ivanovic> so it seems as if cmake is used especially by packagers
20150405 07:13:54< Ivanovic> ;)
20150405 07:14:44< shadowm> IIRC in Debian's case at least, somebody got frustrated early on with scons' unrealiability regarding system config changes between runs (and the need to use --config=force when that happens).
20150405 07:15:20< shadowm> That was back in 2008 and it seems like scons is generally less unreliable these days, but there are things that still don't work the way they should, like out of tree builds.
20150405 07:16:21< iceiceice> i think its honestly pretty bizarre the set up we have
20150405 07:16:25< iceiceice> where the devs use one build system
20150405 07:16:32< iceiceice> and the packagers use a completely different one
20150405 07:16:37< iceiceice> for the most part
20150405 07:16:37< shadowm> It required a lot of experimentation and back and forth correspondence with loonycyborg to get that done for the server, when with cmake I'd have been able to do it in a matter of minutes.
20150405 07:17:29< shadowm> Well, I don't think it's any weirder than Windows-based developers using software that packagers can't possibly use on Linux for obvious reasons.
20150405 07:18:47< iceiceice> its also bad i think that the cmake and scons scripts really have diverged at the level of logic it seems
20150405 07:18:56< iceiceice> different options,
20150405 07:18:57< iceiceice> different checks,
20150405 07:19:01< iceiceice> different versions required for libs
20150405 07:19:03< shadowm> Ivanovic: When copying the po files from 1.12 over to master, is it enough to do that or do I also have to mess with doc/man and doc/manual?
20150405 07:19:27< Ivanovic> shadowm: i always do this already when running the po updates
20150405 07:19:32< iceiceice> i doubt if the files are in the same libs also when it is built
20150405 07:19:34< Ivanovic> you should not have to do it for a release
20150405 07:19:48< Ivanovic> unless you want to run an update
20150405 07:19:58< shadowm> I already ran a pot-update on master.
20150405 07:20:01< Ivanovic> and then you should just copy the files and run the normal "you received an update" procedure
20150405 07:20:14< shadowm> But then I found this email and thought I did it wrong because you mention copying files between branches: https://mail.gna.org/public/wesnoth-i18n/2012-08/msg00001.html
20150405 07:20:20< Ivanovic> so plain copying over does not work
20150405 07:20:38< Ivanovic> with this i meant that I apply the files as update to both branches
20150405 07:20:49< Ivanovic> so that they are merged against the respective pot files
20150405 07:21:00-!- [Relic] [~Relic]@2602:306:33a3:6d30:45b9:a8f7:1a3f:71bd] has quit [Quit: Leaving]
20150405 07:21:11< shadowm> What about missed updates, though?
20150405 07:21:19< Ivanovic> missed updates?
20150405 07:21:34< shadowm> For example, say on 1.12 we had a string "foo bar" that wasn't on master prior to the last pot-update.
20150405 07:21:45< Ivanovic> as mentioned: when copying files, always run the full "update the po file against the pot file" procedure
20150405 07:21:50< shadowm> That string is translated in 1.12 and everything but in master it couldn't be because it didn't exist.
20150405 07:21:58< Ivanovic> bad luck
20150405 07:22:12< Ivanovic> easy as that
20150405 07:22:26< Ivanovic> once you receive the next po update for that language, it will be fixed
20150405 07:22:43< shadowm> So, I should just let you and translators update the catalogues on master over time, got it. :)
20150405 07:22:51< Ivanovic> that is: usually sometime when it made sense i did copy over the stable files again into dev
20150405 07:23:04< Ivanovic> but only for the languages where I was sure that noone was already working on dev
20150405 07:23:11< iceiceice> Ivanovic: i did send an email to the list a few months ago which implicitly dealt with cmake: https://mail.gna.org/public/wesnoth-dev/2014-12/msg00022.html
20150405 07:23:22< iceiceice> we agreed that boost filesystem would be mandatory on master,
20150405 07:23:28< iceiceice> but i did not try to change this over in cmake, only in scons
20150405 07:23:52< shadowm> CTRL+F cmake comes up with nothing there.
20150405 07:25:03< shadowm> Ivanovic: Ugh, that wasn't the email I meant to link. It's this one: https://mail.gna.org/public/wesnoth-i18n/2012-08/msg00003.html
20150405 07:25:06< Ivanovic> btw building dev is currently broken for me due to some warning issues in building the lua part
20150405 07:25:41< Ivanovic> shadowm: for me "copy between branches" means: "copy and update against latest pot"
20150405 07:26:05< shadowm> Yeah, and I missed that because I hadn't seen that email before.
20150405 07:26:17< iceiceice> gn
20150405 07:26:19-!- iceiceice [~chris@wesnoth/developer/iceiceice] has quit [Quit: Leaving]
20150405 07:26:37< shadowm> So, should I do that?
20150405 07:52:31< Ivanovic> as long as you make sure to merge against the pot file afterwards: sure
20150405 07:52:46< Ivanovic> okay, 1.12 build fine, dev failed due to a warning in lua
20150405 07:55:39-!- kex [~kex@78.157.29.160] has quit [Remote host closed the connection]
20150405 08:01:01< shadowm> Ivanovic: Can cmake find fribidi for you there, on master?
20150405 08:01:59< shadowm> Also, could you paste the warning(s) about Lua?
20150405 08:03:49< shadowm> (And what compiler + version are you using?)
20150405 08:07:23-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev
20150405 08:18:50< irker477> wesnoth: Ignacio R. Morelle wesnoth:master 92052e1506b7 / / (1529 files in 29 dirs): Translation resync with 1.12 + pot and documentation update take 2 http://git.io/vec0p
20150405 08:19:00< shadowm> IT IS DONE.
20150405 08:19:03< shadowm> Let us never speak of this again.
20150405 08:19:28-!- Coffee_irc [~david@ppp121-45-36-236.lns20.adl2.internode.on.net] has quit [Quit: Konversation terminated!]
20150405 08:32:09-!- zookeeper [~lmsnie@wesnoth/developer/zookeeper] has joined #wesnoth-dev
20150405 08:34:21< shadowm> shikadibot: seen Randypk
20150405 08:34:21< shikadibot> shadowm: The person with the nick randypk 4d 5h ago they left with the message: Quit: Page closed
20150405 08:35:52< shadowm> gfgtdf: Are you going to handle https://github.com/wesnoth/wesnoth/pull/387 ? If so, please ask this person to add themselves to Miscellaneous Contributors in data/core/about.cfg
20150405 08:54:58-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Quit: Konversation terminated!]
20150405 09:08:41-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev
20150405 09:16:19-!- Haudegen [~quassel@85.124.51.57] has quit [Ping timeout: 255 seconds]
20150405 09:30:01-!- Haudegen [~quassel@85.124.51.57] has joined #wesnoth-dev
20150405 09:47:11-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Ping timeout: 252 seconds]
20150405 10:02:11-!- tomreyn [~tomreyn@megaglest/team/tomreyn] has joined #wesnoth-dev
20150405 10:14:36-!- cib0 [~cib@p5DD215B4.dip0.t-ipconnect.de] has joined #wesnoth-dev
20150405 10:22:09-!- cib0 [~cib@p5DD215B4.dip0.t-ipconnect.de] has quit [Remote host closed the connection]
20150405 11:20:41-!- irker477 [~irker@uruz.ai0867.net] has quit [Quit: transmission timeout]
20150405 11:20:57-!- Kwandulin [~Miranda@p5B00890E.dip0.t-ipconnect.de] has quit [Read error: Connection reset by peer]
20150405 11:40:26-!- prkc [~prkc@4E5CD609.dsl.pool.telekom.hu] has joined #wesnoth-dev
20150405 11:55:04-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev
20150405 12:09:06-!- Haudegen [~quassel@85.124.51.57] has quit [Ping timeout: 250 seconds]
20150405 12:10:09-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Ping timeout: 250 seconds]
20150405 12:21:59-!- Haudegen [~quassel@85.124.51.57] has joined #wesnoth-dev
20150405 12:43:18-!- Kwandulin [~Miranda@p5B00890E.dip0.t-ipconnect.de] has joined #wesnoth-dev
20150405 13:03:43-!- stikonas [~gentoo@wesnoth/translator/stikonas] has joined #wesnoth-dev
20150405 13:08:26-!- ancestral [~ancestral@174-20-187-184.mpls.qwest.net] has joined #wesnoth-dev
20150405 13:14:04-!- gfgtdf [~chatzilla@f054143211.adsl.alicedsl.de] has joined #wesnoth-dev
20150405 13:14:40< gfgtdf> shadowm: i wasnt planning on handle it.
20150405 13:14:56< gfgtdf> shadowm: i also dont know how that fix works
20150405 13:16:28-!- gfgtdf [~chatzilla@f054143211.adsl.alicedsl.de] has quit [Client Quit]
20150405 13:25:14-!- ancestral [~ancestral@174-20-187-184.mpls.qwest.net] has quit [Quit: i go nstuf kthxbai]
20150405 13:33:00-!- kex [~kex@78.157.29.160] has joined #wesnoth-dev
20150405 13:37:26-!- trewe [~trewe@2001:8a0:d108:a601:626c:66ff:fe92:9b7c] has joined #wesnoth-dev
20150405 13:46:09-!- cib0 [~cib@p5DD215B4.dip0.t-ipconnect.de] has joined #wesnoth-dev
20150405 14:23:48-!- iwaim__ [~iwaim@2001:2c0:40e:2002:0:4:14:80] has quit [Ping timeout: 252 seconds]
20150405 14:24:12-!- iwaim__ [~iwaim@2001:2c0:40e:2002:0:4:14:80] has joined #wesnoth-dev
20150405 14:48:49-!- [Relic] [~Relic]@2602:306:33a3:6d30:29ae:a05a:1958:432c] has joined #wesnoth-dev
20150405 14:53:12-!- gfgtdf [~chatzilla@f054143211.adsl.alicedsl.de] has joined #wesnoth-dev
20150405 14:57:18-!- Appleman1234 [~Appleman1@122.103.208.108] has joined #wesnoth-dev
20150405 15:00:07< Ravana_> I would guess that based on archived one this is not supposed to be shared
20150405 15:01:20< gfgtdf> ah yes
20150405 15:02:53< gfgtdf> i forgot that there is a proivacy field, and somehow assumed that servity=security automaticly implies private.
20150405 15:07:08< gfgtdf> shadowm: maybe we can delet tht one from mail archieve ?
20150405 15:11:00-!- cib0 [~cib@p5DD215B4.dip0.t-ipconnect.de] has quit [Remote host closed the connection]
20150405 15:15:13< gfgtdf> iceiceice: ma made a but report for lua print function: https://gna.org/bugs/index.php?23442
20150405 15:30:19-!- shadowm_desktop [ignacio@wesnoth/developer/shadowm] has joined #wesnoth-dev
20150405 15:31:24< shadowm> gfgtdf: Once an email is out, it can't be deleted.
20150405 15:31:28-!- mjs-de [~mjs-de@p508C93F6.dip0.t-ipconnect.de] has joined #wesnoth-dev
20150405 15:31:43< shadowm> Now stop spreading the link around, please.
20150405 15:32:28< shadowm> This goes to Ravana as well.
--- Log opened Sun Apr 05 15:33:16 2015
20150405 15:33:21< Ravana_> the one I showed is the one marked private anyways
20150405 15:33:42< shadowm> Yes, but it has the number in it, which makes it trivial to search for it on the Web.
20150405 15:40:52-!- iceiceice [~chris@wesnoth/developer/iceiceice] has joined #wesnoth-dev
20150405 15:40:59< iceiceice> gfgdf: do you have stack trace?
20150405 15:41:03< iceiceice> i did not get a segfault when i tested
20150405 15:41:39< iceiceice> it doesn't look entirely normal though
20150405 15:44:16< gfgtdf> iceiceice: its teh same error as shadowm reported yesteday
20150405 15:44:35< gfgtdf> iceiceice: it see s like msvc gives a sefault but some other compiler throw an exception
20150405 15:44:43< iceiceice> do you have most recent commit?
20150405 15:44:47< iceiceice> https://github.com/wesnoth/wesnoth/commit/91440ba4f385cadb6d91510634218e133bbe58c4
20150405 15:44:53-!- new_one [~new_one@50.15.126.83] has joined #wesnoth-dev
20150405 15:46:21< gfgtdf> iceiceice: no i doidnt i'll try
20150405 15:48:26< gfgtdf> iceiceice: actually from loking the te code i'm quite sure it'll work
20150405 15:50:01< gfgtdf> iceiceice: hm wait
20150405 15:50:07< gfgtdf> iceiceice: i think i shoudl still test
20150405 15:50:47< iceiceice> yeah it still behaves slightly strange for me, not sure why right now
20150405 15:50:53< iceiceice> but i don't crash either
20150405 15:52:08< gfgtdf> iceiceice: your log object accept NULL cstrings ?
20150405 15:52:11< gfgtdf> our*
20150405 15:52:55< iceiceice> i mean i thought it should just do what string stream will do
20150405 15:53:17< iceiceice> is it illegal to push null into a stream?
20150405 15:53:37< gfgtdf> iceiceice: my guess would be yes but im not sure.
20150405 15:53:57< iceiceice> hmm ok
20150405 15:56:52< shadowm> FTR `char* test = nullptr; std::cerr << "bar\n"; std::cerr << test << "foo" << '\n';` only prints "bar".
20150405 15:57:13< iceiceice> gfgtdf: i think you are right
20150405 15:57:18< shadowm> So it's a bad idea to do so at least.
20150405 15:57:24< iceiceice> i think its undefined behavior
20150405 15:59:11-!- Appleman1234 [~Appleman1@122.103.208.108] has quit [Ping timeout: 256 seconds]
20150405 16:00:10< gfgtdf> Is it normal that 'git fetch upstream master' wants to download 2422 'objects' ? seems to be a lot to me.
20150405 16:00:18< gfgtdf> you know what those 'objects' are ?
20150405 16:01:30< shadowm> Three pot-updates in a row.
20150405 16:03:26-!- irker934 [~irker@uruz.ai0867.net] has joined #wesnoth-dev
20150405 16:03:26< irker934> wesnoth: Chris Beck wesnoth:master 53eac42935e8 / src/scripting/lua_kernel_base.hpp: fix lua logging when lua gives a null string http://git.io/veWUw
20150405 16:04:45< iceiceice> gfgtdf: ^
20150405 16:04:58< iceiceice> it seems to work correctly for me now
20150405 16:05:24< gfgtdf> iceiceice: maybe this line can still cause problems: https://github.com/wesnoth/wesnoth/blob/master/src/scripting/lua_kernel_base.cpp#L94 ?
20150405 16:05:32< gfgtdf> iceiceice: lod you getst with log debug lua ?
20150405 16:05:35< gfgtdf> did*
20150405 16:08:53< iceiceice> yeah i see
20150405 16:08:57< iceiceice> i still dont get segfault but the stream dies
20150405 16:31:51< irker934> wesnoth: Chris Beck wesnoth:master 7a870f016a1c / src/scripting/lua_kernel_base.cpp: fixup lua print with null strings http://git.io/veWYa
20150405 16:31:56< iceiceice> gfgtdf: ^
20150405 16:43:26-!- ancestral [~ancestral@174-20-187-184.mpls.qwest.net] has joined #wesnoth-dev
20150405 16:45:01< loonycyborg> iceiceice: scons works fine even if I run it from a dir with unicode chars, via a symlink
20150405 16:45:33< iceiceice> loonycyborg: it went back to working for me after i renamed my dir to ascii chars
20150405 16:46:05< loonycyborg> I know but still, I can't repro that non-ascii issue
20150405 16:46:59< iceiceice> idk what i can do, do you think that the specific characters matter?
20150405 16:48:09< iceiceice> i also dont know exactly what the error message meant:
20150405 16:48:09< iceiceice> TypeError: unicode argument expected, got 'str'
20150405 16:48:15< iceiceice> http://pastebin.com/GQkS5pPS
20150405 16:50:53< iceiceice> i typed python --version,
20150405 16:50:54< iceiceice> python --version
20150405 16:50:54< iceiceice> Python 2.7.6
20150405 16:51:03< iceiceice> should i have python 3.x.x?
20150405 16:54:00-!- cib0 [~cib@p5DD215B4.dip0.t-ipconnect.de] has joined #wesnoth-dev
20150405 17:00:11-!- Haudegen [~quassel@85.124.51.57] has quit [Ping timeout: 246 seconds]
20150405 17:02:13< iceiceice> loonycyborg: this guy seems to say that the %s syntax doesn't handle unicode in python 2:
20150405 17:02:13< iceiceice> http://stackoverflow.com/questions/13674663/pythons-string-format-and-unicode
20150405 17:03:49-!- gfgtdf [~chatzilla@f054143211.adsl.alicedsl.de] has quit [Ping timeout: 248 seconds]
20150405 17:06:43-!- iceiceice [~chris@wesnoth/developer/iceiceice] has quit [Quit: Leaving]
20150405 17:08:03-!- Kwandulin [~Miranda@p5B00890E.dip0.t-ipconnect.de] has quit [Read error: Connection reset by peer]
20150405 17:14:45-!- Haudegen [~quassel@85.124.51.57] has joined #wesnoth-dev
20150405 17:17:26-!- gfgtdf [~chatzilla@f054143211.adsl.alicedsl.de] has joined #wesnoth-dev
20150405 17:47:03< gfgtdf> iceiceice: commits looks fine
20150405 17:57:33< shadowm> ancestral: Would you be able to do 1.12.2 this Wednesday?
20150405 17:58:02< ancestral> What time?
20150405 17:58:16< shadowm> Unspecified.
20150405 17:58:55< ancestral> On Wednesday, or anytime before the end of day Wednesday?
20150405 17:59:05< ancestral> e.g. today, tomorrow…
20150405 17:59:19< shadowm> On Wednesday, UTC.
20150405 17:59:25< ancestral> (trying not to be pedantic; I work odd hours this week)
20150405 17:59:47< shadowm> Bearing in mind that packages are expected to be available within the next 72 hours before the announcement.
20150405 18:00:16< ancestral> Sorry, I’m not sure I understand
20150405 18:00:20< ancestral> When would the announcement be?
20150405 18:00:36< shadowm> At most 72 hours after tagging on Wednesday.
20150405 18:01:59< ancestral> Tuesday night (starting at 8 PM CDT) or Wendesday night (starting at 9 PM CDT) would be the options I have
20150405 18:02:06< ancestral> I do not have time during the day either days
20150405 18:02:30< ancestral> (1.12 build worked fine for me last I tried; 1.13 I’m getting errors, reached out to mattsc abotu that)
20150405 18:02:35< shadowm> What offset is CDT?
20150405 18:02:43< ancestral> -5 or -6, I forget
20150405 18:03:36< ancestral> Otherwise, I will have time on Thursday until maybe 3 PM
20150405 18:03:55< shadowm> -6, it seems, -5 with DST.
20150405 18:04:18< ancestral> Yeah, 5 hours
20150405 18:04:25< shadowm> I don't really care about the exact time as long as it gets made within 72 hours after tagging.
20150405 18:04:52< shadowm> Odds are I'll do the release during the afternoon UTC-03.
20150405 18:06:00< shadowm> Also, don't forget about the test build of master.
20150405 18:06:20< ancestral> That is where I’m getting errors. I PM’d mattsc.
20150405 18:07:16< ancestral> https://paste.ee/p/JKCZW
20150405 18:07:59< shadowm> Don't build ana.
20150405 18:08:42< shadowm> It breaks a number of important things and has gone unmaintained for two development series (why do we even keep it around I don't know).
20150405 18:08:53< ancestral> Well, that’s good to know
20150405 18:10:01-!- shadowm_desktop [ignacio@wesnoth/developer/shadowm] has quit [Ping timeout: 264 seconds]
20150405 18:12:21-!- shadowm_desktop [ignacio@wesnoth/developer/shadowm] has joined #wesnoth-dev
20150405 18:15:28< shadowm> Or one development series and a half, I guess. I forgoet 1.9.x took really long and 1.11.x took even longer.
20150405 18:16:15< ancestral> So, remove /ana, and network_manager_ana.hpp…
20150405 18:16:42< ancestral> and .cpp
20150405 18:18:19< ancestral> network_ana.cpp
20150405 18:18:48-!- kex [~kex@78.157.29.160] has quit [Remote host closed the connection]
20150405 18:19:21-!- kex [~kex@78.157.29.160] has joined #wesnoth-dev
20150405 18:23:55-!- kex [~kex@78.157.29.160] has quit [Ping timeout: 265 seconds]
20150405 18:28:55-!- travis-ci [~travis-ci@ec2-54-89-16-174.compute-1.amazonaws.com] has joined #wesnoth-dev
20150405 18:28:56< travis-ci> gfgtdf/wesnoth-old#447 (config_reload_sp - 295e1bd : gfgtdf): The build passed.
20150405 18:28:56< travis-ci> Build details : http://travis-ci.org/gfgtdf/wesnoth-old/builds/57260374
20150405 18:28:56-!- travis-ci [~travis-ci@ec2-54-89-16-174.compute-1.amazonaws.com] has left #wesnoth-dev []
20150405 18:38:06-!- travis-ci [~travis-ci@ec2-54-81-170-81.compute-1.amazonaws.com] has joined #wesnoth-dev
20150405 18:38:07< travis-ci> gfgtdf/wesnoth-old#448 (master - fd3e69b : gfgtdf): The build is still failing.
20150405 18:38:07< travis-ci> Build details : http://travis-ci.org/gfgtdf/wesnoth-old/builds/57260566
20150405 18:38:07-!- travis-ci [~travis-ci@ec2-54-81-170-81.compute-1.amazonaws.com] has left #wesnoth-dev []
20150405 18:47:05-!- RiftWalker [Rift-Walke@DHCP-129-59-24-128.n1.vanderbilt.edu] has joined #wesnoth-dev
20150405 18:47:26-!- kex [~kex@78.157.29.160] has joined #wesnoth-dev
20150405 18:48:04-!- noy [~Noy@wesnoth/developer/noy] has quit [Quit: noy]
20150405 18:49:34-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 18:50:39-!- gfgtdf [~chatzilla@f054143211.adsl.alicedsl.de] has quit [Ping timeout: 245 seconds]
20150405 18:52:13-!- kex [~kex@78.157.29.160] has quit [Ping timeout: 264 seconds]
20150405 18:54:36-!- noy [~Noy@wesnoth/developer/noy] has quit [Quit: noy]
20150405 19:21:24-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 19:21:44-!- noy [~Noy@wesnoth/developer/noy] has quit [Client Quit]
20150405 19:22:25-!- mattsc [~mattsc@wesnoth/developer/mattsc] has joined #wesnoth-dev
20150405 19:23:23< mattsc> ancestral: hi — I really don’t have time to be here right now (or much at all today, in fact), just wanted to say hello quickly and that I do not know why you are even seeing any of those errors.
20150405 19:23:42< ancestral> I’ll do another slash and burn
20150405 19:23:45< mattsc> Neither ana nor SDL_gpl appear in the Wesnot Xcode project file and it should never try to build them.
20150405 19:24:05< ancestral> But they’re in master, no?
20150405 19:24:08< ancestral> Hmm
20150405 19:24:31< mattsc> There are some “problems” with new files that have been added recently (as I mentioned previously), but these are not them.
20150405 19:24:38< ancestral> mattsc: I had to add in extras, I may have copied them into it by mistake
20150405 19:24:59< mattsc> okay - that’s always possible
20150405 19:25:10< mattsc> I find myself doing things with Xcode that I did not intend all the time  :P
20150405 19:25:25< ancestral> Right ;-)
20150405 19:25:43< mattsc> If you want me to, I can fix up the Xcode build (in about an hour or so).  Then you know that it works in general, and can troubleshoot on your own side only.
20150405 19:26:45< mattsc> Btw, I did a clean about 10 min ago and then a rebuild.  It’s almost done 450/561 files or so, and I have encountered none of those errors.
20150405 19:27:15< ancestral> Only if it’s not too much trouble for you
20150405 19:27:37< mattsc> ancestral: just to be sure we’re talking about the same thing, you’re build 1.13 as a Debug build, right?
20150405 19:27:43< mattsc> *building
20150405 19:28:16< mattsc> It should be pretty simple, I just can’t do it right this moment.
20150405 19:28:51< mattsc> Yeah - build done, failing in linking mode with an error about countdown_clock  (which is what it should be doing).
20150405 19:29:10< ancestral> Look at that, release not debug. That could make all the difference, eh?
20150405 19:29:21< mattsc> Uh, yes ...
20150405 19:29:52< mattsc> The 1.13 release build needs *a lot* of attention (as I also mentioned ;) )
20150405 19:30:40< ancestral> Still seeing SDL errors (master build)
20150405 19:30:49< mattsc> ancestral: just as a side comment, make sure that you also do a test package, not just a build.  There’s a bunch of things in that process that can go wrong too (they’re easy once you figure them out, but you know how it goes).
20150405 19:30:51< ancestral> I’ll slash and burn and give it a go
20150405 19:30:57< mattsc> sounds good
20150405 19:31:03< ancestral> Sure
20150405 19:31:06< mattsc> I’ll check back in again in about an hour or so
20150405 19:31:11< ancestral> Okay
20150405 19:31:16< mattsc> good luck!
20150405 19:31:18-!- mattsc [~mattsc@wesnoth/developer/mattsc] has quit [Quit: Not here much atm. Will check logs for my nick and forum PMs.]
20150405 19:31:20< ancestral> Thanks
20150405 19:33:09-!- irker934 [~irker@uruz.ai0867.net] has quit [Quit: transmission timeout]
20150405 19:43:31-!- gfgtdf [~chatzilla@f054143211.adsl.alicedsl.de] has joined #wesnoth-dev
20150405 19:45:57-!- kex [~kex@78.157.9.170] has joined #wesnoth-dev
20150405 19:46:23-!- kex [~kex@78.157.9.170] has quit [Read error: Connection reset by peer]
20150405 19:46:41-!- kex [~kex@78.157.9.170] has joined #wesnoth-dev
20150405 19:47:00-!- kex [~kex@78.157.9.170] has quit [Remote host closed the connection]
20150405 19:47:07-!- kex [~kex@78.157.9.170] has joined #wesnoth-dev
20150405 20:02:49-!- shadowm_desktop [ignacio@wesnoth/developer/shadowm] has quit [Ping timeout: 264 seconds]
20150405 20:08:28< gfgtdf> vultraz: whart you mean in https://github.com/wesnoth/wesnoth/pull/245  ? cannot find any comments from shadowm there
20150405 20:10:20< vultraz> gfgtdf: agh, I was unclear. iceiceice's paste of something shadowm said on the bug in question :P
20150405 20:10:22< vultraz> not the pr
20150405 20:11:19-!- cib0 [~cib@p5DD215B4.dip0.t-ipconnect.de] has quit [Remote host closed the connection]
20150405 20:11:33< gfgtdf> vultraz: since there doesnt seem to be a progress i think i should do it, any opinion?
20150405 20:11:53< vultraz> do what?
20150405 20:14:27< gfgtdf> implemeting optional acceleration in [delay]
20150405 20:16:12< vultraz> if have a certain key, this delay will respect acceleration settings?
20150405 20:16:16< vultraz> I guess, if you want
20150405 20:16:28< vultraz> but definitely a false default
20150405 20:16:31-!- mattsc [~mattsc@wesnoth/developer/mattsc] has joined #wesnoth-dev
20150405 20:16:31-!- irker921 [~irker@uruz.ai0867.net] has joined #wesnoth-dev
20150405 20:16:31< irker921> wesnoth: mattsc wesnoth:master b16f9225ebd2 / projectfiles/Xcode/Wesnoth.xcodeproj/project.pbxproj: Update Xcode project file http://git.io/velZz
20150405 20:17:00< mattsc> ancestral: ^
20150405 20:17:11< ancestral> Cool
20150405 20:17:19< ancestral> 80% receiving objects
20150405 20:17:30< ancestral> from git clone
20150405 20:17:33< mattsc> The 1.13 debug build _should_ now work “out of the box”.
20150405 20:17:55< ancestral> Thanks, I appreciate it
20150405 20:18:07< mattsc> As in, if you get a new download of the repository and the “Mac compile stuff”, you should not need anything else.
20150405 20:18:09< gfgtdf> vultraz: so teh key sould be use_acceleration ?
20150405 20:18:10< ancestral> I realize it’s just getting the right files in the project
20150405 20:18:28< ancestral> mattsc: Is the file pango.aliases required/present?
20150405 20:19:05< mattsc> Well, but the zip file should contain everything you need; you should not have to download/add anything else
20150405 20:19:26< mattsc> emphasis is on “should”
20150405 20:19:40< vultraz> gfgtdf: why not just accelerate=
20150405 20:20:29< mattsc> ancestral: just checked, I don’t have pango.aliases in my project directory
20150405 20:20:56< ancestral> Does that get generated during the build process?
20150405 20:21:03< ancestral> If you don’t know, no worries
20150405 20:21:24< ancestral> Somehow that’s come up, I just don’t remember the context :-\
20150405 20:21:42< gfgtdf> vultraz: yes that sounds better, you think that also ok if the prefereces are actually set to a deceleration?
20150405 20:21:44< ancestral> I won’t worry about it then
20150405 20:22:33< mattsc> ancestral: just searched through the Build dir; it’s not in there either in 1.13, so it looks like I got rid of it when I updated pango and cairo.
20150405 20:22:44< vultraz> gfgtdf: yeah, sounds good
20150405 20:23:14< gfgtdf> vultraz: we let's call it accelerate=
20150405 20:27:23-!- mjs-de [~mjs-de@p508C93F6.dip0.t-ipconnect.de] has quit [Remote host closed the connection]
20150405 20:31:12< gfgtdf> vultraz: any opinion: https://github.com/gfgtdf/wesnoth-old/commit/4b34bc4cfc641a8e22288f4c518dbacf3fd3ccee ?
20150405 20:31:40< mattsc> ancestral: I have to go again and won’t be back until evening my time at the earilest.
20150405 20:32:07< ancestral> No problem. Thank you
20150405 20:33:01< gfgtdf> iceiceice: you know when game_display_ in lua_game_kernel coudl be NULL ?
20150405 20:33:05< vultraz> gfgtdf: Add should be capitalized in the changelog 
20150405 20:33:38< mattsc> ancestral: I’ll check the logs and my PMs later.  Hope it’ll work for you without too much trouble from here on.
20150405 20:33:49< ancestral> That’s my hope too
20150405 20:34:25< gfgtdf> iceiceice: I mena if game_display_ is not, wouldnt is usually make more sense to return immidiately ?
20150405 20:34:55< mattsc> If you find anything that you do need to add (theoretically it’s possible that my Xcode uses a non-standard system file, although I tried to make sure that that’s not the case), please write that down so that we can add it to the Mac compile stuff later.
20150405 20:37:12< gfgtdf> vultraz: fixed ver4sion: https://github.com/gfgtdf/wesnoth-old/commit/b622dde96078f48403737f089bc3ea68bbcba706
20150405 20:37:14< mattsc> Anyways, good luck and talk to you later.
20150405 20:37:17-!- mattsc [~mattsc@wesnoth/developer/mattsc] has quit [Quit: Not here much atm. Will check logs for my nick and forum PMs.]
20150405 20:39:36< vultraz> gfgtdf: what does  game_display_->turbo_speed() return if acceleration is set to 1 (default) or disabled
20150405 20:39:56< gfgtdf> vultraz: i think it return1 in both cases
20150405 20:42:10< vultraz> gfgtdf: so accelerate = yes with no acceleration enabled or on default speed would have no different from acclerate=no right?
20150405 20:42:19< gfgtdf> vultraz: yes
20150405 20:42:24< vultraz> ok, looks good then
20150405 20:57:43< gfgtdf> iceiceice: you know why dofile uses lua_call and require uses protected_call ?
20150405 21:02:55-!- travis-ci [~travis-ci@ec2-54-80-166-53.compute-1.amazonaws.com] has joined #wesnoth-dev
20150405 21:02:56< travis-ci> gfgtdf/wesnoth-old#449 (linger_fix - 4b34bc4 : gfgtdf): The build has errored.
20150405 21:02:56< travis-ci> Build details : http://travis-ci.org/gfgtdf/wesnoth-old/builds/57272763
20150405 21:02:56-!- travis-ci [~travis-ci@ec2-54-80-166-53.compute-1.amazonaws.com] has left #wesnoth-dev []
20150405 21:03:10-!- RiftWalker [Rift-Walke@DHCP-129-59-24-128.n1.vanderbilt.edu] has quit [Read error: Connection reset by peer]
20150405 21:09:34-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 21:15:24-!- noy [~Noy@wesnoth/developer/noy] has quit [Quit: noy]
20150405 21:26:23-!- kex [~kex@78.157.9.170] has quit [Remote host closed the connection]
20150405 21:27:43-!- kex [~kex@78.157.9.170] has joined #wesnoth-dev
20150405 21:28:35-!- kex [~kex@78.157.9.170] has quit [Read error: Connection reset by peer]
20150405 21:29:00-!- kex [~kex@78.157.9.170] has joined #wesnoth-dev
20150405 21:31:29-!- travis-ci [~travis-ci@ec2-54-80-166-53.compute-1.amazonaws.com] has joined #wesnoth-dev
20150405 21:31:30< travis-ci> gfgtdf/wesnoth-old#450 (linger_fix - b622dde : gfgtdf): The build passed.
20150405 21:31:30< travis-ci> Build details : http://travis-ci.org/gfgtdf/wesnoth-old/builds/57273404
20150405 21:31:30-!- travis-ci [~travis-ci@ec2-54-80-166-53.compute-1.amazonaws.com] has left #wesnoth-dev []
20150405 21:36:45-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 21:37:20-!- kex [~kex@78.157.9.170] has quit [Remote host closed the connection]
20150405 21:38:57-!- noy [~Noy@wesnoth/developer/noy] has quit [Client Quit]
20150405 21:47:07-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 21:50:06-!- noy [~Noy@wesnoth/developer/noy] has quit [Client Quit]
20150405 21:51:15-!- kex [~kex@78.157.9.170] has joined #wesnoth-dev
20150405 21:55:03-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 21:56:36-!- noy [~Noy@wesnoth/developer/noy] has quit [Client Quit]
20150405 22:10:03-!- kex [~kex@78.157.9.170] has quit [Remote host closed the connection]
20150405 22:16:38-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 22:16:57-!- noy [~Noy@wesnoth/developer/noy] has quit [Client Quit]
20150405 22:27:19-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 22:29:12-!- noy [~Noy@wesnoth/developer/noy] has quit [Client Quit]
20150405 22:29:57-!- stikonas [~gentoo@wesnoth/translator/stikonas] has quit [Quit: Konversation terminated!]
20150405 22:31:12-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 22:33:23-!- noy [~Noy@wesnoth/developer/noy] has quit [Client Quit]
20150405 22:36:03-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 22:36:03-!- noy [~Noy@wesnoth/developer/noy] has quit [Client Quit]
20150405 22:37:50-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 22:46:15-!- noy [~Noy@wesnoth/developer/noy] has quit [Quit: noy]
20150405 22:46:59-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 22:48:01-!- noy [~Noy@wesnoth/developer/noy] has quit [Client Quit]
20150405 22:48:18-!- anoob [~anoob@189-30-87-247.jvece702.dsl.brasiltelecom.net.br] has joined #wesnoth-dev
20150405 22:49:16< ancestral> 1.12 builds fine for me, 1.13 builds fine, but I get a CTD
20150405 22:49:21< anoob> is it possible to speed up builds? now i need to compile the whole wesnoth source every time i change something ...
20150405 22:49:33< irker921> wesnoth: gfgtdf wesnoth:master cd0029eac19a / changelog data/lua/wml-tags.lua src/scripting/game_lua_kernel.cpp: add accelerate in [delay] http://git.io/vel5T
20150405 22:49:49< gfgtdf> iceiceice: ^
20150405 22:50:07-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 22:50:27-!- noy [~Noy@wesnoth/developer/noy] has quit [Client Quit]
20150405 22:50:54< gfgtdf> iceiceice: any opinion? otherwise i'd close https://github.com/wesnoth/wesnoth/pull/245.
20150405 22:51:13< gfgtdf> ancestral: what is CDT ?
20150405 22:51:26< ancestral> Crash to Desktop
20150405 22:51:40< ancestral> CDT is Central Daylight Time ;-)
20150405 22:51:43< gfgtdf> anoob: usually you just need to recompile the changes files (unless you changes a header that is oncluded in every file)
20150405 22:52:09< gfgtdf> ancestral: any error message or stacktrace ?
20150405 22:52:16< ancestral> Yes
20150405 22:52:39< gfgtdf> anoob: usually that depends on which compiler etc you use
20150405 22:52:45< ancestral> https://paste.ee/p/lWiBo
20150405 22:52:52< gfgtdf> anoob: it's betst to get help from soneone who used teh same compiler
20150405 22:53:33< ancestral> afaik mattsc doesn’t get this crash, so I’m guessing it is something wrong with how I built it
20150405 22:55:31< anoob> i always run scons again and again ... does it detect just what has changed?
20150405 22:56:30< gfgtdf> ancestral: maybe you nned a later verison of libpango?
20150405 22:56:33< gfgtdf> need*
20150405 22:56:45< ancestral> Perhaps!
20150405 22:57:05< ancestral> I used the 1.13 compile files on sourceforge. I could try newer pango
20150405 22:58:01< gfgtdf> ancestral: are that teh same files as in 1.12 ?
20150405 22:58:13< ancestral> I don’t think so
20150405 22:58:24< ancestral> The ones in 1.12 are the 1.11 compile files
20150405 22:58:55< gfgtdf> ancestral: maybe the 1.12 files are better maintained ?
20150405 22:58:59< ancestral> The next thing I’ll try is the 1.11 building for 1.13, just to see
20150405 22:59:01< ancestral> Yeah
20150405 22:59:24< gfgtdf> ancestral: i think its strange that we have different files for 1.12 and 1.13
20150405 22:59:51< gfgtdf> ancestral: or it it maybe becasue 1.12 want to support oder OS versions but 1.13 not ?
20150405 22:59:58< gfgtdf> ancestral: it worked?
20150405 23:00:10< ancestral> gfgtdf: So I think how it works is with a new dev branch we’ve updated the versions of the libs
20150405 23:00:18< ancestral> And that carries over to the next release
20150405 23:00:52-!- Appleman1234 [~Appleman1@s108.BMT-e1.vectant.ne.jp] has joined #wesnoth-dev
20150405 23:02:17-!- noy [~Noy@wesnoth/developer/noy] has joined #wesnoth-dev
20150405 23:03:01< gfgtdf> ancestral: do we have some type of guarantee that teh dependencies do not change during a stable version ?
20150405 23:04:28< ancestral> My understanding is yes. So when we build 1.12.2 it will still have the same version of those files
20150405 23:05:04-!- Appleman1234 [~Appleman1@s108.BMT-e1.vectant.ne.jp] has left #wesnoth-dev []
20150405 23:05:07< ancestral> mattsc could probably better answer these questions, and for what it’s worth, I’ve pinged him about the crash-to-desktop
20150405 23:05:59< gfgtdf> ancestral: it worked when you used the 1.12 files ?
20150405 23:07:01-!- noy [~Noy@wesnoth/developer/noy] has quit [Ping timeout: 264 seconds]
20150405 23:07:03< ancestral> 1.12 works with the 1.11/1.12 compile files
20150405 23:07:13< ancestral> Let me try 1.13 with the 1.11/1.12 compile files
20150405 23:07:17-!- iceiceice [~chris@wesnoth/developer/iceiceice] has joined #wesnoth-dev
20150405 23:07:25< iceiceice> gfgtdf: i looked at the server code about making a message related to this: https://gna.org/bugs/index.php?22936
20150405 23:08:06< iceiceice> it's fairly difficult because we don't track the version number after they log in
20150405 23:08:12< gfgtdf> iceiceice: y i know
20150405 23:08:38< iceiceice> i think it might be easier if
20150405 23:08:42< iceiceice> we just dont make a server message
20150405 23:08:52< iceiceice> and advise mp campaigns to put a {VERSION} tag in their code or somethign
20150405 23:09:19< ancestral> Fails right away
20150405 23:09:21< ancestral> During the build
20150405 23:09:50< gfgtdf> iceiceice: how would a version tag fix this ?
20150405 23:10:16< iceiceice> they could print a warning if the wesnoth version is below 1.12.1
20150405 23:10:29< gfgtdf> iceiceice: ah i see
20150405 23:10:40< iceiceice> there's not actually that many mp campaigns
20150405 23:11:02< iceiceice> i mean we could print a message when they log in
20150405 23:11:13< iceiceice> "if you were thinking to play mp campaigns... dont do that..." or something
20150405 23:11:49< iceiceice> but i think its probably easier if its done in wml
20150405 23:12:02< iceiceice> also idk if shadowm decided there were security problems in 1.12.1 or not
20150405 23:12:20< iceiceice> we could just have announcement login "you are recommended to upgrade to 1.12.2 to fix various bugs"
20150405 23:12:25< gfgtdf> iceiceice: there were
20150405 23:13:37< zookeeper> gfgtdf, the add-on is Scenario_with_robots, but at least _main.cfg uses Scenario_with_Robots in include paths
20150405 23:14:12< iceiceice> gfgtdf: i'm not going to try to make a server message on campaign transition,
20150405 23:14:19< iceiceice> just put something in release notes
20150405 23:14:35< gfgtdf> zookeeper: ty :)
20150405 23:18:13-!- zookeeper [~lmsnie@wesnoth/developer/zookeeper] has quit [Ping timeout: 248 seconds]
20150405 23:22:17-!- shadowm_desktop [ignacio@wesnoth/developer/shadowm] has joined #wesnoth-dev
20150405 23:22:20< shadowm> iceiceice: 1.11.19, 1.12.0, 1.12.1.
20150405 23:22:24< gfgtdf> iceiceice: so you now wont make rthe {version} macro ?
20150405 23:22:50< shadowm> 1.10.0 through 1.10.7 are likely affected as well.
20150405 23:23:38< gfgtdf> shadowm: didnt you say its impossible to delete from the mail log ?
20150405 23:27:38-!- shadowm_desktop [ignacio@wesnoth/developer/shadowm] has quit [Ping timeout: 244 seconds]
20150405 23:28:54< shadowm> I did. Then a couple of requests were processed.
20150405 23:29:25-!- shadowm_desktop [ignacio@wesnoth/developer/shadowm] has joined #wesnoth-dev
20150405 23:31:38< irker921> wesnoth: Chris Beck wesnoth:1.12 5b02203d5162 / RELEASE_NOTES: update RELEASE_NOTES http://git.io/velxM
20150405 23:31:58-!- mattsc [~mattsc@wesnoth/developer/mattsc] has joined #wesnoth-dev
20150405 23:32:31< mattsc> ancestral: hi — another one of my here-for-a-few-minutes visit  :|
20150405 23:32:45< gfgtdf> iceiceice: and opinion about http://git.io/vel5T ?
20150405 23:32:50< mattsc> ancestral: how do you start the executable?
20150405 23:32:57< ancestral> Double clicing from the Finder
20150405 23:33:02< ancestral> *clicking
20150405 23:33:06< mattsc> Could you start from the commandline and let me know what you see int he terminal?
20150405 23:33:10< ancestral> sure
20150405 23:33:47< mattsc> gfgtdf, ancestral: the reason why we have different libraries is because the once used by 1.12 are way old.
20150405 23:34:11< mattsc> I updated version of pango, cairo, boost, … and also went to fat binaries that support both 32 and 64 bit builds.
20150405 23:34:54< mattsc> Unlike Linux, Xcode does not do this (using the latest libraries) automatically.  The hope is that homebrew will eventually take care of this.
20150405 23:35:04< ancestral> Gotta rebuild
20150405 23:35:15< shadowm> Linux doesn't do it automatically either. ;)
20150405 23:35:18< iceiceice> gfgtdf: looks good to me i think
20150405 23:35:23< iceiceice> i didnt think about the units
20150405 23:35:35< mattsc> well, there’s a CL command that does it for you, or something like that ...
20150405 23:35:38< iceiceice> i guess that the luainteger is supposed to be time in ms ?
20150405 23:35:45< gfgtdf> iceiceice: i'll then close https://github.com/wesnoth/wesnoth/pull/245.
20150405 23:35:46< iceiceice> lua integer is not a double or something, right?
20150405 23:35:48< iceiceice> yeah sure
20150405 23:36:14< gfgtdf> iceiceice: lua int is teh return type of luaL_checkinteger
20150405 23:36:16< shadowm> It just tends to be easier to obtain the latest libraries, especially on source code-based distributions like Gentoo (I think), and then there's the Linux equivalent of homebrew.
20150405 23:36:33< gfgtdf> iceiceice: its signed int 64 but thatmigth be plattform sependent
20150405 23:36:35< mattsc> ancestral: and yes, the 1.13 build does start for me without a crash.  I actually did try after I updated the Xcode project file earlier.
20150405 23:36:38< shadowm> The latter isn't ubiquitous I think, seeing as how I only heard of it last year and I've been using this platform for 10 years.
20150405 23:37:49< iceiceice> gfgtdf: i read the patch again, it looks good to me
20150405 23:38:39< gfgtdf> iceiceice: ok
20150405 23:39:06< gfgtdf> iceiceice: you know when game_display_ cal be null ?
20150405 23:39:10< gfgtdf> can*
20150405 23:39:43< iceiceice> if we ever could make game state copyable,
20150405 23:39:53< iceiceice> then i guess you would want to detach the game_display from the copy
20150405 23:40:11< shadowm> gfgtdf: Don't forget to add [delay] accelerate to the wiki with {{DevFeature1.13|0}}: http://wiki.wesnoth.org/InterfaceActionsWML#.5Bdelay.5D
20150405 23:40:18< mattsc> ancestral: btw, I just tried starting from Finder (I usually start from the CL so that I have stdout right there) and no crash for that either.
20150405 23:40:59< ancestral> Just for the heck of it, I tried the 1.12 headers and libs and it failed to build. So I’m replacing them with the 1.13’s again
20150405 23:41:25< mattsc> ancestral: yes, that will definitely not work
20150405 23:41:42< ancestral> *1.11’s I mean
20150405 23:41:52< ancestral> Alright, recompiling
20150405 23:42:46< gfgtdf> shadowmdone
20150405 23:43:26< gfgtdf> iceiceice: but during that phase (while detached) we usually dont rexecute lua/wml right ?
20150405 23:43:53< iceiceice> i mean we could eventually
20150405 23:43:57< iceiceice> it doesnt exist right now
20150405 23:44:16< iceiceice> i think there might be another time that game_display_ is null
20150405 23:44:37< gfgtdf> iceiceice: when ?
20150405 23:44:48< gfgtdf> iceiceice: is there a wiki page on ignore file pbl ?
20150405 23:45:02< shadowm> http://wiki.wesnoth.org/IGNFileFormat ?
20150405 23:45:18< shadowm> Eek, no not that page.
20150405 23:45:24< ancestral> 320/564
20150405 23:45:25< iceiceice> gfgtdf: i think that game display is null during the game setup 
20150405 23:45:37< iceiceice> have to check to be sure
20150405 23:46:04< shadowm> Well, actually, that *is* the right page. I seemed to remember it being longer.
20150405 23:46:54< gfgtdf> shadowm: ty
20150405 23:46:57< iceiceice> gfgtdf: right so, in theory the whole gamestate should get built here:
20150405 23:46:58< iceiceice> https://github.com/wesnoth/wesnoth/blob/master/src/play_controller.cpp#L212
20150405 23:47:11< iceiceice> that includes map, unit map, teams, 
20150405 23:47:21< iceiceice> event queue, lua stuff,
20150405 23:47:27< gfgtdf> shadowm:  think there shoudl be link to it in http://wiki.wesnoth.org/PblWML
20150405 23:47:54< iceiceice> the whiteboard and display get attached here: https://github.com/wesnoth/wesnoth/blob/master/src/play_controller.cpp#L249
20150405 23:49:44< iceiceice> it might be briefly detached also whent he replay gets restarted and everything gets torn down and reconstructed
20150405 23:49:54< gfgtdf> shadowm: in windows in ignfile do i use \ or / dor deirecotry seperators ?
20150405 23:50:00< shadowm> I've added a link to See Also for now because I don't really feel like writing documentation right now.
20150405 23:50:17< gfgtdf> shadowm: see also if fine
20150405 23:50:23< ancestral> mattsc: OK
20150405 23:50:30< ancestral> Segmentation fault: 11
20150405 23:50:47< shadowm> gfgtdf: You may only use /, \ won't do anything.
20150405 23:50:48< ancestral> https://paste.ee/p/31C7P
20150405 23:50:48< mattsc> anything before that?
20150405 23:50:54< ancestral> Yes
20150405 23:51:13< mattsc> Ugh
20150405 23:51:19< mattsc> That’s not very helpful ...
20150405 23:51:21< ancestral> That’s right after “Battle for Wesnoth v1.13.0-dev”
20150405 23:51:29< ancestral> “Started on Sun Apr  5 18:50:02 2015”
20150405 23:51:30< ancestral> I know
20150405 23:51:49< ancestral> I put the crash log up online
20150405 23:52:00< mattsc> yeah, I saw it
20150405 23:52:12< ancestral> https://paste.ee/p/lWiBo
20150405 23:52:15< ancestral> k
20150405 23:52:17< mattsc> unfortunately, I am not used to reading those things.  I didn’t get anything out of that.
20150405 23:52:28< ancestral> For sure
20150405 23:52:37< ancestral> I did my best and couldn’t find any nuggets
20150405 23:52:39< shadowm> ancestral: --log-debug=all should give you all the output before the crash.
20150405 23:52:50< ancestral> shadowm: thanks, let me try that
20150405 23:52:58< gfgtdf> shadowm: the default .*/ already implies the .git/ directory right ?
20150405 23:53:05< mattsc> shadowm: yes, right …  Thanks
20150405 23:53:18< ancestral> Fantastic
20150405 23:53:23< shadowm> It sounds to me like it's running in GUI2 code prior to the crash, while trying to render text via Cairo/Pango.
20150405 23:54:06< shadowm> On the GUI2 title screen, to be specific.
20150405 23:54:43< shadowm> gfgtdf: .* matches all dot directories, so yes.
20150405 23:55:13< ancestral> OK it’s long
20150405 23:55:17< ancestral> This is the last part
20150405 23:55:17< ancestral> https://paste.ee/p/8Ov4V
20150405 23:56:37< ancestral> Let me try a bigger chunk
20150405 23:57:02< ancestral> mattsc, shadowm: https://paste.ee/p/shRKO
20150405 23:57:29< ancestral> mattsc: You know what I’d be curious in trying?
20150405 23:57:55< ancestral> Running your binary. It’s possible I may be playing a little with fire. I’m on 10.10.3 beta
20150405 23:58:23< ancestral> If your binary were to crash on my computer, then we’ll know I need to be running 10.10.2 or something else
20150405 23:58:25< mattsc> ancestral: okay - I’ve been too lazy to upgrade to Mavericks so far …
20150405 23:58:30< ancestral> Aha
20150405 23:58:51< mattsc> But I am using the 10.10 SDKs in Xcode
20150405 23:58:59< ancestral> Right, and 1.12 works
20150405 23:59:20< ancestral> mattsc: For the target, I was choosing 10.6 — that wouldn’t cause an issue after build, I assume?
20150405 23:59:35< mattsc> Umm, yes, that might be the problem.
20150405 23:59:45< mattsc> Try 10.8 - let me check though what I am using first ...
20150405 23:59:47< ancestral> What are you setting the target at for 1.13?
20150405 23:59:49< ancestral> Okay
--- Log closed Mon Apr 06 00:00:05 2015