Re: [Bf-committers] [Bf-blender-cvs] SVN commit: /data/svn/bf-blender [45556] trunk/blender/CMakeLists.txt: cmake/mingw now builds with default configuration (needed to disable tiff/ exr/cycles)
Yet another commit that cripples MinGW. MinGW builds fine with gcc 4.6.2 and all of the above libraries enabled. The cmake command line I use here is: cmake -G MinGW Makefiles -DWITH_RAYOPTIMIZATION=ON -DWITH_CODEC_FFMPEG=ON -DWITH_PLAYER=ON -DWITH_IMAGE_OPENEXR=ON -DWITH_OPENCOLLADA=ON -DWITH_CYCLES=ON -DWITH_CYCLES_CUDA_BINARIES=ON -DWITH_MOD_OCEANSIM=ON -DWITH_FFTW3=ON -DWITH_LIBMV=ON ..\blender And it just works (with CUDA binaries MSVC is required due to a CUDA compiler dependency). Is this based on user's input? I really can't reproduce a non-building setup on two systems (Win 7 and WIn XP) Can we have this reverted please? ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] build blender with mingw
Hi, fighting with some build issues today and remembered this thread. Since builders complaining about MinGW not building is an often recurring theme, it is important to know that only gcc 4.6.2 is supported. Use the latest mingw-get-inst and download latest packages. That should take care of the dependency. Also, if you want to build CUDA binaries, it is unfortunately a requirement of the CUDA compiler that MSVC is installed too, and you need to run the build from a MSVC command line. At least that is the case with the 4.0 toolkit. On MinGW64 issues, I am going to investigate further. A poster on BA gave this link http://mamedev.org/tools/ where it is clear that MAME use their own distribution of MinGW64 too. It is very rational IMO, for the reasons I have posted previously and should we support MinGW64 we should also distribute our own version. I just wish there was a debug version of python for MinGW and CUDA didn't require MSVC for building. Then we could really have a standalone compiler system with MinGW(64?) ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] [Bf-blender-cvs] SVN commit: /data/svn/bf-blender [45561] trunk/blender/source/blender/ blenkernel/BKE_global.h: add in a check for Mingw so 4. 5 or older will error out on build.
I'm not actually fan of such changes. Why to force me to use latest mingw version to test my areas which compiles fine with gcc-4.4? I still maintaining FFmpeg on windows which still can not be compiled by gcc newer than 4.4 because of some crappy things going on with sse and optimizations. Can not it be a warning or error if some unsupported things are enabled? On Thu, Apr 12, 2012 at 11:10 AM, Campbell Barton ideasma...@gmail.comwrote: Revision: 45561 http://projects.blender.org/scm/viewvc.php?view=revroot=bf-blenderrevision=45561 Author: campbellbarton Date: 2012-04-12 09:10:10 + (Thu, 12 Apr 2012) Log Message: --- add in a check for Mingw so 4.5 or older will error out on build. Modified Paths: -- trunk/blender/source/blender/blenkernel/BKE_global.h Modified: trunk/blender/source/blender/blenkernel/BKE_global.h === --- trunk/blender/source/blender/blenkernel/BKE_global.h2012-04-12 08:45:26 UTC (rev 45560) +++ trunk/blender/source/blender/blenkernel/BKE_global.h2012-04-12 09:10:10 UTC (rev 45561) @@ -166,6 +166,11 @@ # error Either __BIG_ENDIAN__ or __LITTLE_ENDIAN__ must be defined. #endif +/* there is really no good place for this */ +#if defined(FREE_WINDOWS) ((__GNUC__ 4) || ((__GNUC__ == 4) (__GNUC_MINOR__ 6))) +# error Mingw requires GCC 4.6 minimum +#endif + #define L_ENDIAN 1 #define B_ENDIAN 0 ___ Bf-blender-cvs mailing list bf-blender-...@blender.org http://lists.blender.org/mailman/listinfo/bf-blender-cvs -- With best regards, Sergey Sharybin ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] [Bf-blender-cvs] SVN commit: /data/svn/bf-blender [45561] trunk/blender/source/blender/ blenkernel/BKE_global.h: add in a check for Mingw so 4. 5 or older will error out on build.
Hi Sergey, personally I voted for 4.6 because 4.5 crashed when compiling libmv due to running out of stack. For 4.6 they raised the limit. As far as i know 4.6 can produce a full-featured blender executable. I wasn't aware of the problems with ffmpeg but it gives me no problems when linking the already presentffmpeg version here, so I guess you can keep compiling the library with gcc 4.4. ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] build blender with mingw
On Thu, Apr 12, 2012 at 12:11:24PM +0300, Antony Riakiotakis wrote: On MinGW64 issues, I am going to investigate further. A poster on BA gave this link http://mamedev.org/tools/ where it is clear that MAME use their own distribution of MinGW64 too. It is very rational IMO, for the reasons I have posted previously and should we support MinGW64 we should also distribute our own version. I cannot imagine that we would want to be in the business of distributing MinGW. -- Stephen Swaney sswa...@centurytel.net ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] [Bf-blender-cvs] SVN commit: /data/svn/bf-blender [45561] trunk/blender/source/blender/ blenkernel/BKE_global.h: add in a check for Mingw so 4. 5 or older will error out on build.
On blenderartist, a few users have had a hard time getting blender to build with mingw (after spending a lot of time on this aparently). So seemed reasonable to ensure mingw/cmake work out of the box - Since new devs (*also experienced devs*) can loose a lot of time chasing up cryptic error messages and its not a nice introduction to blender dev. We should at least attempt to give users a reasonable experience building blender, it really sucks when someone tells you PROGRAM/VERSION XYZ is known to fail when you just spent some hrs trying to figure out whats going on. - Disable known non-working configurations. - Disable options that fail with mingw. - Default CMake config should work right away. @Sergey, I had the same issue where GCC crashes in with version 4.5 and works with 4.6, rather then disabling it we could make libMV give some error with mingw and an older GCC. CMake/Scons could be used to do this instead rather then #error however getting compiler version for both CMake and SCons isnt so simple. As it turns out the linking errors I had with cycles seem spesific to my setup - and Im still not sure why cycles/tiff/exr fail. see: http://www.graphicall.org/ftp/ideasman42/mingw_4_6_err.txt On Fri, Apr 13, 2012 at 12:44 AM, Antony Riakiotakis kal...@gmail.com wrote: Hi Sergey, personally I voted for 4.6 because 4.5 crashed when compiling libmv due to running out of stack. For 4.6 they raised the limit. As far as i know 4.6 can produce a full-featured blender executable. I wasn't aware of the problems with ffmpeg but it gives me no problems when linking the already presentffmpeg version here, so I guess you can keep compiling the library with gcc 4.4. ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers -- - Campbell ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] Nodes auto hide sockets on collapsed nodes
Let me heby begin a little battle against auto-hide sockets of collapsed nodes: Please take a look: https://vimeo.com/40237189 Bartek Skorupa www.bartekskorupa.com On 2012-04-10, at 09:47, Ton Roosendaal wrote: Hi Lukas, I would like to see such usability changes in nodes to go via at least the other listed node UI owners too. For me huge alarm bells ring when people talk about adding options or giving users choices. That's usually just patching flawed designs... Remove the autohide for now? We should spend design powers on how we want to work with node systems in general. -Ton- Ton Roosendaal Blender Foundation t...@blender.org www.blender.org Blender Institute Entrepotdok 57A 1018AD Amsterdam The Netherlands On 9 Apr, 2012, at 21:10, François T. wrote: I'm with Bartek on this one :( 2012/4/9 Lukas Tönne lukas.toe...@googlemail.com I missed it because now you select socket by left clicking. I was trying to select them by right clicking (as it was in 2.49) and when it failed I was sure we still don't have this option available. This feature is a real time saver. The primary way of selecting sockets is with SHIFT + RMB. First this was implemented as simple RMB select, but this turned out to be quite chaotic since it effectively allows multiple nodes selected (see [1]). Selecting sockets with LMB is a side effect of the select-on-link feature (which i'd like to remove again). You see this area is pretty complicated, lots of different little features interfering with each other :S Auto hide sockets: As I can see from your post you are all the time thinking about automatic behavior. If I may suggest something - I'd rather leave more options to the user. Yeah, i implemented this on explicit request (i think Sebastian asked for it, don't remember), i suggest you talk to him. The CTRL+H idea seems good, i didn't realize this doesn't work on the auto-hidden sockets now. Feedback is important for us coders :) Will have to check with core devs if i can add this right away of have to wait for next BCon1. [1] http://lists.blender.org/pipermail/bf-blender-cvs/2012-March/044066.html ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers -- François Tarlier www.francois-tarlier.com www.linkedin.com/in/francoistarlier ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] build blender with mingw
Well, if maintaining MinGW is hell already I really can't see how else it can be done :/. -if- it is done... ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] [Bf-blender-cvs] SVN commit: /data/svn/bf-blender [45561] trunk/blender/source/blender/ blenkernel/BKE_global.h: add in a check for Mingw so 4. 5 or older will error out on build.
I didn't experience any issues with libmv in my setup, only cycles gave errors. Ofcourse i can use another mingw to compile blender, but it's kind of crap which i think might be avoided :S On Thu, Apr 12, 2012 at 9:14 PM, Campbell Barton ideasma...@gmail.comwrote: On blenderartist, a few users have had a hard time getting blender to build with mingw (after spending a lot of time on this aparently). So seemed reasonable to ensure mingw/cmake work out of the box - Since new devs (*also experienced devs*) can loose a lot of time chasing up cryptic error messages and its not a nice introduction to blender dev. We should at least attempt to give users a reasonable experience building blender, it really sucks when someone tells you PROGRAM/VERSION XYZ is known to fail when you just spent some hrs trying to figure out whats going on. - Disable known non-working configurations. - Disable options that fail with mingw. - Default CMake config should work right away. @Sergey, I had the same issue where GCC crashes in with version 4.5 and works with 4.6, rather then disabling it we could make libMV give some error with mingw and an older GCC. CMake/Scons could be used to do this instead rather then #error however getting compiler version for both CMake and SCons isnt so simple. As it turns out the linking errors I had with cycles seem spesific to my setup - and Im still not sure why cycles/tiff/exr fail. see: http://www.graphicall.org/ftp/ideasman42/mingw_4_6_err.txt On Fri, Apr 13, 2012 at 12:44 AM, Antony Riakiotakis kal...@gmail.com wrote: Hi Sergey, personally I voted for 4.6 because 4.5 crashed when compiling libmv due to running out of stack. For 4.6 they raised the limit. As far as i know 4.6 can produce a full-featured blender executable. I wasn't aware of the problems with ffmpeg but it gives me no problems when linking the already presentffmpeg version here, so I guess you can keep compiling the library with gcc 4.4. ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers -- - Campbell ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers -- With best regards, Sergey Sharybin ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] API: How to get a list of running scripts?
how about incrementing a global variable each frame? Can a new instance wait one frame and then actually launch if the value of the variable hasn't changed? my scripts are triggered by Redraw not by Frame change, so i have thought about time stamps and delay value checks, but it is not reliable, because the idle periods may be unpredictable long. Anyway, seems like I'm the only one responding to this thread, and I'm in no way an expert... Have you tried to ask main developers in the IRC directly? asked Campbell, and the answer is, 2.49 API doesn't support this. Though it should be easy to extend - i have to look at Blender.Scene.getScriptLinks() method. As I've learned recently, in 2.63 (or maybe even 2.6+), there are bpy.app.handlers.scene_update_pre and bpy.app.handlers.scene_update_post -- these event handlers are invoked each frame (not sure if this behavior will persist, though). It's possible to get the list of the handlers, so I guess in recent Blender versions your idea can be implemented. i am keeping my eyes on 2.6, but awaiting significant moves in UI+API area after bmesh integration at first. ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] Nodes auto hide sockets on collapsed nodes
LOL, nice video and I do think you have a point. On Thu, Apr 12, 2012 at 5:36 PM, Bartek Skorupa bartekskor...@bartekskorupa.com wrote: Let me heby begin a little battle against auto-hide sockets of collapsed nodes: Please take a look: https://vimeo.com/40237189 Bartek Skorupa www.bartekskorupa.com On 2012-04-10, at 09:47, Ton Roosendaal wrote: Hi Lukas, I would like to see such usability changes in nodes to go via at least the other listed node UI owners too. For me huge alarm bells ring when people talk about adding options or giving users choices. That's usually just patching flawed designs... Remove the autohide for now? We should spend design powers on how we want to work with node systems in general. -Ton- Ton Roosendaal Blender Foundation t...@blender.org www.blender.org Blender Institute Entrepotdok 57A 1018AD Amsterdam The Netherlands On 9 Apr, 2012, at 21:10, François T. wrote: I'm with Bartek on this one :( 2012/4/9 Lukas Tönne lukas.toe...@googlemail.com I missed it because now you select socket by left clicking. I was trying to select them by right clicking (as it was in 2.49) and when it failed I was sure we still don't have this option available. This feature is a real time saver. The primary way of selecting sockets is with SHIFT + RMB. First this was implemented as simple RMB select, but this turned out to be quite chaotic since it effectively allows multiple nodes selected (see [1]). Selecting sockets with LMB is a side effect of the select-on-link feature (which i'd like to remove again). You see this area is pretty complicated, lots of different little features interfering with each other :S Auto hide sockets: As I can see from your post you are all the time thinking about automatic behavior. If I may suggest something - I'd rather leave more options to the user. Yeah, i implemented this on explicit request (i think Sebastian asked for it, don't remember), i suggest you talk to him. The CTRL+H idea seems good, i didn't realize this doesn't work on the auto-hidden sockets now. Feedback is important for us coders :) Will have to check with core devs if i can add this right away of have to wait for next BCon1. [1] http://lists.blender.org/pipermail/bf-blender-cvs/2012-March/044066.html ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers -- François Tarlier www.francois-tarlier.com www.linkedin.com/in/francoistarlier ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers -- Douglas E Knapp Creative Commons Film Group, Helping people make open source movies with open source software! http://douglas.bespin.org/CommonsFilmGroup/phpBB3/index.php Massage in Gelsenkirchen-Buer: http://douglas.bespin.org/tcm/ztab1.htm Please link to me and trade links with me! Open Source Sci-Fi mmoRPG Game project. http://sf-journey-creations.wikispot.org/Front_Page http://code.google.com/p/perspectiveproject/ ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] Nodes auto hide sockets on collapsed nodes
I agree with Bartek, and having some kind of hover over socket and the name of the socket is somehow displayed would be amazing. -NateW ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] API: How to get a list of running scripts?
Hi, If this is about an addon, there seems to be a function to check if it is running: import addon_utils addon_utils.check('space_view3d_3d_navigation') If not then it seems more difficult. I might be wrong here, but I think in that case there isn't really any running 'script object' that exists. There might be python objects in memory that were created by the script, like functions that were installed as handlers somewhere, but it's just bits of data attached in various places. An error in the code will not remove those python objects, they stick around until you actually go and delete them. Brecht. On Wed, Apr 11, 2012 at 3:40 AM, Remigiusz Fiedler mig...@gmx.net wrote: hi all, I am looking for a method for conditional start of a script only if no instance of it is running already. How to get a list of running scripts through API? thanks in advance, migius ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers
Re: [Bf-committers] Nodes auto hide sockets on collapsed nodes
your video made me laugth :) +1 2012/4/12 Nate Wiebe natewieb...@gmail.com I agree with Bartek, and having some kind of hover over socket and the name of the socket is somehow displayed would be amazing. -NateW ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers -- François Tarlier www.francois-tarlier.com www.linkedin.com/in/francoistarlier ___ Bf-committers mailing list Bf-committers@blender.org http://lists.blender.org/mailman/listinfo/bf-committers