--- Log opened Sun Feb 25 00:00:23 2007 01:15 -!- madsen__ is now known as madsen 01:17 < madsen> anyone experienced (just a tiny bit) in building synfig on OS X? 01:18 * madsen isn't the slightest bit experienced in building anything on OS X, but desperately wants synfig back. 01:21 -!- crazy_bus [n=philip@60.229.247.180] has joined #synfig 03:01 -!- crazy_bus [n=philip@60.229.247.180] has quit [Read error: 104 (Connection reset by peer)] 03:01 -!- dyloxin [n=dyloxin@dsl081-054-146.sfo1.dsl.speakeasy.net] has quit [Read error: 104 (Connection reset by peer)] 03:14 < dooglus> madsen: still here? 03:15 < madsen> yeah 03:15 < dooglus> ChrisVP was here earlier asking the same question 03:15 < dooglus> I tried helping him build it (I've built in under Linux, but never used OS X) 03:16 < madsen> dooglus: Hehe, I've been using Debian for 7 years and switched to OS X 3 weeks ago... 03:16 < dooglus> we got part way through, but then got stuck with incompatibilities: 03:16 < dooglus> (22:17:59) ChrisVP: i think i'm going to leave this and come back later 03:16 < dooglus> (22:18:24) ChrisVP: there's just too much to handle with the fink/macports conflicts and the dependencies and everything else 03:16 < dooglus> madsen: how have you found OS X compared to Debian? 03:17 < madsen> dooglus: Well, OS X is shiny, but I lack control. 03:17 < madsen> Especially I miss apt/dpkg (and no, Fink just isn't the same - I want it to take care of my entire system). 03:17 < dooglus> right 03:18 < dooglus> I think I would miss apt too 03:18 < dooglus> I occasionally use Windows and it seems crazy the amount of work I have to do to keep apps up to date 03:18 < madsen> indeed! 03:19 < dooglus> have you been able to install g++, libgtkmm and all the other dependencies on OS X? 03:19 < madsen> Most OS X apps can check for updates themselves, but I like a centralized system for that. 03:19 < madsen> dooglus: Well, I installed xcode dev tools, so that much, yeah. The rest I installed with Fink. 03:19 < dooglus> http://wiki.synfig.com/Build_instructions lists the dependencies 03:19 < madsen> g++ comes with xcode. 03:20 < madsen> I was able to install everything for synfig (not synfigstudio) with fink - except from libltdl, which is optional. 03:20 < dooglus> the svn repository has files called 'macosxbuild.sh' in the top level - is that any use? 03:20 < madsen> Only problem is that I can't compile synfig for some weird reason. I talked to pabs about it last night. 03:21 < dooglus> ChrisVP was OK with synfig, too, but got stuck on synfigstudio's deps I think 03:21 < madsen> Well, that's what I tried - and it seems to work, but I seem to lack something on my system. 03:21 < madsen> He managed to build synfig with all the optional build deps? 03:22 < madsen> I can configure it, but once make starts, it dies. 03:22 < dooglus> well, I'm not sure how well it built... 03:22 < madsen> some problem with isnan/isnanf 03:22 < madsen> Hmm... 03:23 < dooglus> there's a log of our conversation about it here: http://dooglus.rincevent.net/random/2007-02-24.204503+0100CET.txt 03:23 < madsen> pabs' and my convo is in the logs from yesterday - if you want to see. 03:24 < madsen> Dang, can't get used to have to use Command-C to copy from a terminal. I miss the select (=copy) and middle-click (=paste) from Linux. 03:27 < madsen> Hmm, it seems he actually didn't manage to build synfig (core) after all... Doesn't it? 03:28 < dooglus> what I don't understand is the "(22:13:31) ChrisVP: i don't have any of that" line... if he didn't have any of the source at all, how did he build anything? 03:28 < dooglus> have you used synfig a lot on debian? 03:28 < madsen> Maybe he was looking in the wrong dir? 03:28 < madsen> dooglus: Well, I only got to use it for 2 weeks, then my Thinkpad died and I bought a Mac. 03:28 < dooglus> ok 03:28 < dooglus> I've been going through the open bugs in the tracker, trying to fix them 03:29 < madsen> Oh! Yeah, I wish I could help fix them, but I only know enough C/C++ to say "hello world" (and ask where the toilet is :-p ). 03:29 < dooglus> it's tricky, since I don't know of any documentation that says how things should work... 03:29 < madsen> hehe 03:30 < madsen> Yeah, it would be nice to standardize Synfig a bit. 03:30 < madsen> In the build process it seems that 'make' calls ./configure before actually doing the 'make'. 03:34 -!- crazy_bus [n=philip@60.229.247.180] has joined #synfig 03:36 < madsen> I really wish that some Mac-guru would drop by and reveal to me all the secrets of building stuff for OS X. 03:36 < dooglus> the 'make' only runs 'configure' if it needs to, I guess 03:36 < madsen> dooglus: Nope, it just runs it as far as I can see. 03:37 < dooglus> I think you would be better off finding an OS X channel, rather than waiting for a Mac guru to arrive here... 03:37 < madsen> Yeah, you're probably right. Or maybe the fink-channel... 03:38 < madsen> ... though, I can only imagine them being pretty tired of os x-noobs crying for help. 03:40 < madsen> Personally, I'd prefer to build an app-bundle, so people could just drag n' drop it to install. You know, with all the required libs included and stuff... Maybe a .pkg, if a bundle isn't possible. 03:40 < madsen> Is there any way to figure tell synfig where to look for libs? 03:40 < madsen> And any way of figuring out what files should be included in before-mentioned bundle? 03:41 < madsen> I mean, what libs synfig calls... 03:42 < dooglus> well, the debian package lists its dependencies, if that's what you mean 03:42 < dooglus> $ apt-cache show synfig | grep Depends 03:42 < dooglus> Depends: libc6 (>= 2.3.6-6), libgcc1 (>= 1:4.1.0), libglib2.0-0 (>= 2.10.0), libglibmm-2.4-1c2a, libsigc++-2.0-0c2a (>= 2.0.2), libstdc++6 (>= 4.1.0), libsynfig0, libxml++2.6-2, libxml2 (>= 2.6.26) 03:43 < madsen> Yeah, but I reckon libc6 (for instance) shouldn't be included in the OS X app-bundle. 03:43 < madsen> That should be present on OS X already. 03:43 < madsen> I mean what libs are called dynamically from the binary and where it finds them. 03:44 < dooglus> I guess you need a clean install of OS X to test your bundle on 03:44 < madsen> May be... 03:44 < madsen> I have plenty of non-dev type friends, who have nothing funky installed - they could test it. 03:45 < dooglus> I think pretty much everything is linked dynamically 03:45 < madsen> ... and they use OS X, that is. On PPC though, but a Universal binary would be preferable anyway. 03:45 < madsen> dooglus: Yeah, then it should look for some .so or .o files somewhere. But which and where? 03:46 < madsen> And can I tell it where? 03:47 < madsen> Problem is, that an app-bundle is basically just a dir-structure disguised as a single file on OS X... It allows for inclusion of libraries and other resources along with the binary/binaries, I just need to make the binary look for it in that dir-structure and not in, e.g., /usr/lib or something. 03:48 < dooglus> in Linux, it uses the LD_LIBRARY_PATH environment variable to know where to look for .so files 03:48 < dooglus> and I can run "readelf -d /home/chris/programs/synfig/synfig/src/tool/.libs/synfig | grep NEEDED" to see which libraries it's going to look for 03:48 < dooglus> it says: 03:48 < dooglus> 0x00000001 (NEEDED) Shared library: [libsynfig.so.0] 03:48 < dooglus> 0x00000001 (NEEDED) Shared library: [libxml++-2.6.so.2] 03:48 < dooglus> 0x00000001 (NEEDED) Shared library: [libxml2.so.2] 03:48 < dooglus> 0x00000001 (NEEDED) Shared library: [libglibmm-2.4.so.1] 03:48 < dooglus> 0x00000001 (NEEDED) Shared library: [libgobject-2.0.so.0] 03:48 < dooglus> 0x00000001 (NEEDED) Shared library: [libglib-2.0.so.0] 03:48 < dooglus> 0x00000001 (NEEDED) Shared library: [libsigc-2.0.so.0] 03:48 < dooglus> 0x00000001 (NEEDED) Shared library: [libpthread.so.0] 03:48 < dooglus> 0x00000001 (NEEDED) Shared library: [libstdc++.so.6] 03:48 < dooglus> 0x00000001 (NEEDED) Shared library: [libm.so.6] 03:48 < dooglus> 0x00000001 (NEEDED) Shared library: [libgcc_s.so.1] 03:48 < dooglus> 0x00000001 (NEEDED) Shared library: [libc.so.6] 03:49 < madsen> Now where getting somewhere! :) 03:50 < madsen> And if LD_LIBRARY_PATH contains more paths, it'll prefer libs in the first-mentioned paths over the other? 03:52 < dooglus> I *think* LD_LIBRARY_PATH doesn't need to be set at all, and it will look wherever the libraries were when the program was linked, and then in /usr/lib/ 03:52 < dooglus> but if it is set, it'll look through it (it's colon-separated) in order, just like how 'PATH' is used to find executables 03:53 < dooglus> (the above are just for 'synfig' by the way - for 'synfigstudio' I run readelf -d /home/chris/programs/synfig/synfigstudio/src/gtkmm/.libs/lt-synfigstudio | grep NEEDED - and get an even longer list of libraries 03:53 < madsen> Hmm... But that's not too good on OS X, because the libraries only will be found by a relative path. (From the binary it'll be ../Framework/.) 03:53 < dooglus> (34 of them) 03:54 < madsen> Well, they can be included in the bundle - but will only the .so files be needed? I reckon I'll have to compile all of them for mac - intel as well as ppc. 03:54 < dooglus> I imagine that you would link using "-L../Framework", and then it'll check ../Framework at run time 03:54 < madsen> Ok, that's doable. 03:54 < dooglus> (but I'm guessing) 03:55 < dooglus> is this how things generally work on OS X? every app ships a complete copy of the libs it requires? 03:55 < madsen> I'm starting to think there should be some xcode way to do this - after all, it is the native OS X dev-tool. 03:55 < dooglus> what happens when a bug is fixed in one of the libs? you have to re-install all the apps which use it? 03:55 < madsen> Well, I think so - if you want to make the installation simple that is. 03:56 < madsen> But X11 will be required, so we can depend on the common x11 libs to be installed. 03:57 < dooglus> I see things like: 03:57 < dooglus> 0x00000001 (NEEDED) Shared library: [libXinerama.so.1] 03:57 < dooglus> in the dependencies 03:57 < dooglus> can you assume that Xinerama is installed? 03:57 < madsen> hang on a sec, I'll check. 03:58 < madsen> I guess you can - it's in /usr/X11R6/lib, so installed with X11. 03:58 < madsen> I can see, though, that Gimp.app (the OS X version of The Gimp) ships its own though. 03:59 < madsen> /Applications/graphics/Gimp.app/Contents/Resources/lib/libXinerama.1.0.dylib 04:00 < madsen> I was thinking about doing a similar thing, so you'd only have to rely on X11 being installed (either Apple's X11 or one from MacPorts or Fink). 04:02 < madsen> It'd mean that the binary would be far bigger, but it'd be a breeze to install. (Simply drag n' drop SynfigStudio.app to the Applications folder and you're done.) 04:02 < madsen> That's how things are usually done. Then you can uninstall it simply by deleting SynfigStudio.app and everything will be gone. 04:03 < dooglus> that sounds fine 04:03 < dooglus> I put the list of libraries here: http://dooglus.rincevent.net/synfig/dep.txt 04:04 < madsen> Cool, but I'll probably need to make my own, since I probably won't be able to make a build identical to yours. :-/ 04:06 < madsen> But hey, I think I'm getting closer to a working build. If only I could make 'make' actually work... 04:07 < dooglus> did you see the script in the top level directory? 04:07 < madsen> macosxbuild.sh? 04:09 < dooglus> yes 04:09 < dooglus> does that help? 04:10 < dooglus> it sets a few flags and calls make 04:11 < madsen> Indeed. I messed a little around with it, but it definitely helps, especially because it uses options not even mentioned in ./configure --help. :) 04:12 < dooglus> does that get you past the 'nan' problem you mentioned earlier? 04:13 < madsen> Nope, I discovered it with that script. 04:13 < madsen> I've got an idea though. Let me just try something. :) 04:16 < madsen> Actually, if you look at the script, you'll see that PREFIX is set to /Applications/SynfigStudio.app/Contents/synfig, which means it's already trying to build an app-bundle. :) 04:18 < dooglus> in Debian, isnan() is declared in libc6-dev's /usr/include/bits/mathcalls.h 04:18 < dooglus> (which is pulled in by #include ) 04:19 < madsen> yeah, it's included, but it still messes up... Dunno why... 04:21 < madsen> http://rafb.net/p/cwp1X158.html - most recent error. 04:22 < madsen> Arg, sorry, apparently rafb.net/paste doesn't like to wrap. :( 04:23 < madsen> I guess it's actually more of a compiler issue than a lib-issue... But I'm not sure, since I know nothing about real coding and stuff. (Interpreted languages is more my thing for coding.) 04:28 < madsen> Hmm, perhaps http://developer.mozilla.org/en/docs/Mac_OS_X_Build_Prerequisites can teach me (or us) a thing or two... 04:44 < dooglus> I think you need to define __APPLE__ 04:45 < dooglus> synfig/src/synfig/color.h says: 04:45 < dooglus> #ifdef __APPLE__ 04:45 < dooglus> #define isnan __isnanf 04:45 < dooglus> #endif 04:47 < madsen> Hmm... How do I do that? 04:47 < dooglus> perhaps try: ARCH_FLAGS="$ARCH_FLAGS -arch i386 -D__APPLE__" in the script 04:47 < madsen> Just #define __APPLE__ 04:47 < madsen> Hehe, I will! :) 04:47 < dooglus> (just add the -D__APPLE__ bit to the existing arch flags) 04:48 < madsen> I can see, that I should use gcc4 for intel macs, I'll just try to do that as well... Just a sec. 04:49 < madsen> Hehe, it already does. :) 04:49 < madsen> Trying another ./configure and make. 04:50 < madsen> Nope, same error in the same place. :-/ 04:50 < dooglus> do you see your -D__APPLE__ in the big long g++ line though? 04:51 < madsen> Yup! 04:51 < madsen> g++ -L/Applications/SynfigStudio.app/Contents/synfig/lib -I/Applications/SynfigStudio.app/Contents/synfig/include -D__APPLE__ -arch i386 -DHAVE_CONFIG_H ... 04:53 < madsen> Weird thing is (mentioned this to pabs also) that ./configure gives me: 04:53 < madsen> checking for isnan... yes 04:53 < madsen> checking for isnanf... no 04:53 < madsen> checking for _isnanf... no 04:55 < dooglus> hmm 04:55 < madsen> Here's the config summary - apparently I've got no support for any image types (even though the libs are installed). Don't know if you can use it for anything. http://rafb.net/p/GKktzs50.html 04:56 < dooglus> what do you see in config.h matching 'NAN'? 04:56 < madsen> where is config.h? 04:57 < madsen> or rather, which... 04:57 < madsen> madsen$ find . -iname "config.h" 04:57 < madsen> ./macosxbuild/config.h 04:57 < madsen> ./macosxbuild/libltdl/config.h 04:57 < madsen> ./synfig.pbproj/config.h 04:57 < dooglus> in synfig 04:57 < dooglus> oh, ok, in macosxbuild 04:57 < madsen> K, hang on. 04:58 < madsen> http://rafb.net/p/DzVdwr16.html 04:59 * madsen hugs his nopaste ruby-script. 05:00 < dooglus> can you edit synfig/src/synfig/color.h, around line 56 to look like this: 05:00 < dooglus> #ifdef __APPLE__ 05:00 < dooglus> #warning apple is defined 05:00 < dooglus> #define isnan __isnanf 05:00 < dooglus> #else 05:00 < dooglus> #warning apple not defined 05:00 < dooglus> #endif 05:00 < madsen> Sure! 05:01 < dooglus> then check which of the 2 warnings you see when you build 05:02 < madsen> I assume I should look for the warning somewhere near the error, right? 05:03 < dooglus> yes 05:03 < dooglus> like: In file included from ../../../src/synfig/module.h:36, from main.cpp:33: 05:03 < dooglus> ../../../src/synfig/color.h:60:2: warning: #warning apple not defined 05:04 < madsen> In file included from ../../../src/synfig/gradient.h:31, 05:04 < madsen> from ../../../src/synfig/gradient.cpp:32: 05:04 < madsen> ../../../src/synfig/color.h:57:2: warning: #warning apple is defined 05:05 < madsen> It seems like it tries to use __isnanf because it's apple, but it's not there... But I'm just guessing... 05:05 < dooglus> but the error message complains about isnan, doesn't it? 05:06 < dooglus> can you paste the whole lot again? 05:06 < dooglus> the compile line, and the errors? 05:06 < madsen> sure 05:07 < madsen> http://rafb.net/p/hvwcYJ21.html 05:07 < dooglus> ta 05:07 < madsen> Ach, sorry, now it does C-highlighting on it... :-/ 05:07 < dooglus> ok, the error messase is different 05:08 < madsen> Yeah, that's probably because I had to uncomment the line with #define isnan __isnanf. It was commented out before. 05:08 < madsen> Sorry, I should have mentioned that. :-/ 05:09 < dooglus> is http://www.ssl.berkeley.edu/pipermail/boinc_dev/2004-October/000529.html the same problem? 05:11 < madsen> Might be, but I'm not sure. 05:11 < madsen> Should I try adding the extern "C" { ... stuff in color.h after #include ? 05:12 < dooglus> it's worth a try I suppose 05:12 < madsen> It says that __isnanf wasn't declared in "this" scope. 05:12 < dooglus> (get with of the -D__APPLE__ thing though) 05:12 < madsen> K, I will... 05:13 < dooglus> another page I found suggests: In the newer compiler standard isnan is within std. So replacing replacing 05:13 < dooglus> isnan by std::isnan fixes the problem. 05:14 < dooglus> ( https://mail.gna.org/public/getfem-users/2006-11/msg00007.html ) 05:14 < madsen> Worth a try too! 05:14 < madsen> Perhaps more worth a try, than the ugly hack. 05:15 < dooglus> so: #define isnan std::isnan 05:15 < madsen> in color.h, right? 05:16 < madsen> the 'extern "C"'-thing didn't work. 05:16 < dooglus> in color.h 05:16 < madsen> trying 05:17 < madsen> same error, same place 05:18 < madsen> seems like it's determined to call __isnanf in color.h line 185 (or 184 in the original source). 05:19 < madsen> Guess I could try with the -D__APPLE__ again... It seems to be ignoring the #ifdef __APPLE_... 05:21 < madsen> ... and no luck either... 05:22 < dooglus> before you added the -D__APPLE__ bit, it was saying: ../../../src/synfig/color.h:184: error: 'isnan' was not declared in this scope 05:22 < dooglus> are you sure you've taken the -D__APPLE__ bit out of the script? 05:23 < madsen> yeah, I'm pretty sure - I just put it back in after an unsiccesful make. 05:23 < madsen> the g++ call didn't include -D__APPLE__ either. 05:24 < dooglus> ok 05:26 < madsen> You want the latest error - just to keep you up to date? 05:29 < madsen> http://rafb.net/p/jDLHOC61.html - now wrapped! 05:35 < madsen> Hmm... Maybe I should look into what libc6 I've got... Hang on... 05:37 < madsen> http://lists.cs.uiuc.edu/pipermail/llvmbugs/2004-June/000694.html gave me the idea... 05:37 < dooglus> what if you try compiling this 3-line file, nan.cpp : 05:37 < dooglus> #include 05:37 < dooglus> #include 05:37 < dooglus> using namespace std; int main(void) { cout << isnan(3.14) << endl; } 05:38 < dooglus> just: g++ nan.cpp 05:39 < madsen> compiles and runs (prints 0). 05:40 < madsen> madsen$ g++ nan.cpp -o nan 05:40 < madsen> madsen$ ./nan 05:40 < madsen> 0 05:44 < madsen> Hmm... I have a couple of math.h's lying around... Any way to know which one it uses? 05:46 < madsen> I guess /usr/include/math.h would be the appropriate one, but it's got a heck of a lot to choose from... 05:49 < dooglus> g++ -E nan.cpp will tell you 05:49 < dooglus> pipe it through grep math.h, 'cos it'll be quite verbose 05:51 < dooglus> '0' is the correct output - 3.14 is NOT not a number... 05:51 < madsen> Mmm... Pi...e... 05:52 < dooglus> I think things like 1/0 and 0/0 are what isnan() will be true for 05:52 < madsen> http://rafb.net/p/MSnpaU62.html - output of g++ -E nan.cpp | grep math.h 05:52 < madsen> looks about right to me, but what do I know. 05:54 < dooglus> it means it's using the headers from /usr/include, as you would expect 05:55 < dooglus> you can check the same for the long line that the synfig build is doing 05:55 < madsen> Hmm, that might be interesting... -E is the magic parameter? 05:55 < dooglus> cd into the appropriate directory, and run the long compile line, replacing '-c' with '-E' and removing the '-o' and the following .o file 05:55 < dooglus> -E means 'only run the source through the C preprocessor' - don't compile it 05:55 < madsen> Ok! 05:56 < madsen> Sweet! 05:56 < dooglus> so you get to see what gets #included 05:56 < madsen> Just gotta wait for it to reproduce the error first... 06:00 < madsen> What dir should I cd to? It looks for ../../../src/synfig/value.cpp 06:01 < madsen> ok, src/synfig/proto worked. :) 06:02 < dooglus> you should see lines like: make[3]: Entering directory `/home/chris/programs/synfig/synfigstudio/src/gtkmm' 06:02 < madsen> But it gave output very similar to that of nan.cpp... If not identical. 06:02 < dooglus> they tell you which directory the commands are being run in 06:02 < dooglus> what if you grep the output of the g++ ... -E ... for "nan"? 06:02 < madsen> ok. I'll keep that in mind. :) 06:03 < madsen> the long one? 06:03 < dooglus> well, both I guess 06:03 < dooglus> to see if there's a difference 06:04 < madsen> Hmm, interesting! Something popped up on stderr! (Because I piped stdout to a file this time.) 06:04 < madsen> ../../../src/synfig/value.cpp:29:21: error: config.h: No such file or directory 06:06 < madsen> whoa! There are differences... A LOT! 06:06 < madsen> Well, not that much, but still different... (The diff output just seemed like a lot at first glance.) 06:08 < madsen> Want me to paste the diff output? 06:08 < dooglus> could you upload the whole output of the two g++ -E commands somewhere? 06:09 < madsen> Yeah, I've got a server. You want the whole output or just the grepped stuff? 06:09 < dooglus> the whole lot 06:12 < madsen> Want to see something weird? (In my opinion, at least...) 06:12 < madsen> 618K Feb 25 06:10 nan.cpp.out 06:12 < madsen> 1K Feb 25 06:11 value.cpp.out 06:14 < madsen> http://lillesvin.net/stuff/nan.cpp.out.bz2 and http://lillesvin.net/stuff/value.cpp.out.bz2 06:15 < madsen> Got a bit of a slow upstream, so I hope it's ok I compressed them. 06:17 -!- crazy_bus [n=philip@60.229.247.180] has quit [Remote closed the connection] 06:18 < dooglus> 1K? 06:18 < madsen> Yeah... Odd isn't it? 06:18 < madsen> I think it's because of the missing config.h I mentioned... 06:19 < dooglus> no, it's because you've passed it through 'grep nan'... 06:19 < madsen> Arg! Sorry! 06:20 < dooglus> :) 06:20 -!- pabs3 [n=pabs@dsl-58-7-252-146.wa.westnet.com.au] has joined #synfig 06:20 < dooglus> hey pabs 06:20 < madsen> the command history is a dangerous thing ... apparently. 06:20 < madsen> pabs3: Hey! 06:21 < pabs3> hey folks? 06:21 < madsen> dooglus: 959k better? 06:21 < dooglus> that's more like it 06:22 < madsen> ok, try http://lillesvin.net/stuff/value.cpp.out.bz2 06:22 < dooglus> pabs3: did you see my email? 06:22 < pabs3> yep 06:22 < pabs3> looking at the patch now 06:23 < dooglus> ok 06:23 < dooglus> I just added some tooltips to the icons in the preview window 06:24 < dooglus> because I can never remember which of the two red-blob buttons is 'stop' :) 06:25 < dooglus> madsen: that's from compiling value.cpp, right? 06:25 < dooglus> does value.cpp actually fail to compile with the isnan error? 06:26 < pabs3> :) 06:26 < madsen> yeah, I copy/pasted the last g++ call before the error... I assumed it was the one. 06:26 < pabs3> weird-ass isnan stuff again :/ 06:27 < madsen> pabs3: Yeah, same problem as yesterday - only, this time I'm up even later... And slightly hung over now. :-| 06:27 < pabs3> :) always the best time to be debugging 06:28 < madsen> Hehe... I reckon it's better than throwing around sudo's while drunk. :-p 06:28 < pabs3> hahaha, done that in my time 06:29 < madsen> Yeah, me too... Then I got wiser... Or more forgetful... I can't remember which... 06:30 < pabs3> dooglus: about the loop bug patch thing, as long as it compiles ok, and fixes the bug, I say apply it. if it causes something later down the track we can deal with that then. 06:30 < pabs3> about this patch... 06:30 < dooglus> yes? 06:30 * dooglus hopes he isn't going to be asked to explain how it works 06:31 < madsen> hehe 06:31 < pabs3> well.... 06:32 < dooglus> madsen: could you paste the corresponding terminal session log, showing the compile line and the errors? 06:32 < madsen> dooglus: Sure can, just a sec. 06:32 < dooglus> pabs3: out with it! 06:33 < pabs3> well, I don't really know exactly what it does 06:33 < pabs3> pls explainzor 06:33 < pabs3> :) 06:33 < dooglus> ok 06:33 < madsen> http://rafb.net/p/rs7YUU61.html 06:34 < madsen> dooglus: -^ 06:34 < madsen> dooglus: Oh, sorry, forgot the error, coming up here (only one line). 06:34 < dooglus> you know how there are 2 ways of interactive with a bline: when you draw it with the bline tool, and later, when you edit it with the 'normal' tool 06:34 < pabs3> yep 06:35 < dooglus> the changes to valuenodedynamiclist*loop.cpp are for the 'normal' tool 06:35 < madsen> dooglus: ../../../src/synfig/value.cpp:29:21: error: config.h: No such file or directory (called from src/synfig/proto) 06:35 < dooglus> they make it so you can right-click on the vertices in the canvas to loop and unloop. 06:36 < dooglus> previously you had to go to the param window and right-click the word 'vertices' 06:36 < dooglus> madsen: I'm wanted to see the '-E' output from a command which fails with a 'isnan' type error, not a 'config.h' error - that's a separate issue 06:37 < madsen> dooglus: Oh, so you mean another copy/paste of the error when it compiles? 06:37 < dooglus> pabs3: the patch to state_bline.cpp is a lot simpler - it adds 'unloop bline', and does away with the check that was only making it available for the vertex at the end of the line 06:40 < pabs3> ok, cool. from the email, sounds like you've tested it enough 06:40 < pabs3> so just apply it 06:40 < dooglus> pabs3: yes, and it does seem to work. 06:40 < dooglus> I just don't really understand the details of how! 06:40 < madsen> http://rafb.net/p/WbjlzJ94.html 06:41 < dooglus> all this 'vocab' stuff, and the meaning of 'value_desc's and 'value_node's 06:41 < dooglus> still isn't really clear to me 06:41 < madsen> ... and that was for dooglus as well. 06:41 < pabs3> I'm thinking that as far as reviewing patches goes, synfig-devl is the logical place for that, and if it stuffs something up, thats what svn history is for :) 06:41 < dooglus> ok 06:41 < dooglus> commits go to synfig-devel automatically, right? 06:42 < pabs3> nah 06:42 < dooglus> ok 06:42 < pabs3> I'm looking at putting them on the sidebar on synfig.com, gotta get access to the files on sf.net first 06:48 < dooglus> I've committed the patch and closed the bug 06:48 < dooglus> do you have a working directory you can build it in? 06:49 < madsen> dooglus: pabs or me? 06:50 < dooglus> madsen: pabs3. I know you can't build anything :) 06:50 < pabs3> about the tangent ducks covering the vertex one, any ideas about that? maybe we could merge menus or something 06:50 < madsen> dooglus: Hehe, too true. :-/ 06:50 < dooglus> madsen: as for you... I'm a little confused. are you saying that when you 'make', value.cpp's compilation fails with a 'isnan' error, but when you try running the long g++ line manually it gives a different (config.h) error? 06:50 < pabs3> yep, not very fast computer tho 06:51 < dooglus> pabs3: merging the menus is a reasonable idea 06:52 < madsen> dooglus: yeah... I don't see the config.h error in the make output... 06:52 < dooglus> madsen: I guess that means you're not running it in the right directory, or something 06:53 < madsen> dooglus: Maybe. I'm checking into that as we speak... Just can't find any make-msgs about "entering directory". 06:54 < dooglus> madsen: you're not using GNU make then? 06:54 < dooglus> $ make -v 06:54 < dooglus> GNU Make 3.81 06:54 < dooglus> Copyright (C) 2006 Free Software Foundation, Inc. 06:54 < dooglus> anyway - it's well past my bedtime. 06:54 < madsen> dooglus: madsen$ make -v 06:54 < madsen> GNU Make 3.80 06:54 < madsen> Copyright (C) 2002 Free Software Foundation, Inc. 06:55 < madsen> dooglus: Yeah, me too... Lke, 6 hours past. hehe... 06:55 < dooglus> I'm off now. good luck getting this sorted out. and I'll be back at some point if you need more help. 06:55 < dooglus> heh - you're in Europe as well then? 06:55 < madsen> dooglus: Thanks, most likely I will. 06:55 < madsen> dooglus: Yeah, Denmark. You? 06:56 < pabs3> hmm, I really need to get jhbuild setup for synfig 06:57 < madsen> jhbuild? Not sure I know it... 06:57 < madsen> dooglus: France? 06:57 < pabs3> used by gnome ppl to compile & install cvs sources 06:58 < pabs3> good for testing development versions 06:58 < madsen> Oh! Then I've heard about it. 06:58 < madsen> pabs3: It seems that there are some unorthodox things going on with the build process as well... At least the weird 'make' calls ./configure thing... 06:59 < pabs3> indeed, those wibbly bits need fixing 06:59 * pabs3 wants to murder ./bootstrap 07:00 < madsen> pabs3: Good plan! I'll help you get rid of the corpse. 07:00 < pabs3> :) 07:02 < madsen> ok... this is odd... if I just make ./macosxbuild.sh echo the ./configure call (with all options and flags) and then copy/paste it, then it fails... 07:03 < madsen> if I let the script run it (the exact same line as I copy/paste), then it runs... 07:05 < dooglus> that's because what gets shown on screen is missing the double-quotes 07:05 < madsen> Doh! 07:06 < dooglus> are you using the subversion trunk? 07:06 < madsen> Yeah 07:07 < dooglus> I just committed a fixed synfig/macosxbuild.sh 07:07 < madsen> ok, updating 07:08 < madsen> what'd you change? 07:08 < dooglus> I put backslashes before each " 07:08 < dooglus> to stop the shell eating them 07:08 < madsen> oh! 07:10 < pabs3> woo, dooglus on a roll! 07:11 < madsen> Hmm, just the same. Nothing changes. Same error, same stuff, same everything. I just thought I'd try running it from the root of the working copy and not macosxbuild/, as macosxbuild.sh does. 07:13 < madsen> I better go to bed before I off someone (or something), this isnan problem is driving me nuts! Are we getting any closer to a solution? 07:13 < madsen> I have no clue, 'cause I'm not a C/C++-programmer, and apparently no C/C++-programmers have MacBooks... At least none of those working on synfig. 07:14 < dooglus> I'm too tired to think straight, but next time you're around I'll take a proper look 07:14 < dooglus> goodnight :) 07:14 < madsen> dooglus: Ok. Night! Sleep well! 07:14 < madsen> pabs3: See ya later! 07:14 < pabs3> laters :) 07:15 < madsen> G'night! 07:37 -!- crazy_bus [n=philip@60.229.70.215] has joined #synfig 07:54 -!- crazy_bus [n=philip@60.229.70.215] has quit [Remote closed the connection] 10:59 -!- pabs3 [n=pabs@dsl-58-7-252-146.wa.westnet.com.au] has quit ["Don't rest until all the world is paved in poems."] 11:38 -!- crazy_bus [n=philip@60.229.70.215] has joined #synfig 12:27 -!- crazy_bus [n=philip@60.229.70.215] has quit [Remote closed the connection] 17:24 -!- madsen_ [n=madsen@62.79.53.147] has joined #synfig 17:28 -!- igli [n=igli@82.152.195.63] has joined #synfig 17:38 -!- madsen [n=madsen@62.79.53.147] has quit [Read error: 110 (Connection timed out)] 18:54 -!- vdepizzol [n=vinicius@200.242.12.134] has joined #synfig 19:17 -!- vdepizzol [n=vinicius@200.242.12.134] has quit ["Ex-Chat"] 19:34 -!- vdepizzol [n=vinicius@200.242.12.134] has joined #synfig 21:08 -!- vdepizzol [n=vinicius@200.242.12.134] has quit ["Ex-Chat"] 21:50 -!- omry_ [n=omry@l192-117-99-228.broadband.actcom.net.il] has joined #synfig 22:06 -!- omry [n=omry@l85-130-138-23.broadband.actcom.net.il] has quit [Connection timed out] 22:12 -!- omry_ [n=omry@l192-117-99-228.broadband.actcom.net.il] has quit [Connection timed out] 22:13 -!- omry_ [n=omry@l192-117-123-253.broadband.actcom.net.il] has joined #synfig 22:32 -!- igli [n=igli@82.152.195.63] has quit [Remote closed the connection] 23:06 -!- igli [n=igli@82.152.199.77] has joined #synfig 23:12 -!- madsen_ is now known as madsen 23:19 < dooglus> hey 23:19 < dooglus> any luck with your OS X build? 23:25 < madsen> Nope, haven't tried again since yesterday. 23:26 < madsen> You haven't figured out what kind of dark magic envelopes math.h/isnan on OS X? 23:29 < madsen> Hmm, I gotta go for a while, I'll be back in an hour or so. Later! 23:56 < dooglus> k --- Log closed Mon Feb 26 00:00:23 2007