This week's meeting focused entirely on the Blender 2.70 Release Candidate.
Ton Roosendaal writes:
Hi all,
Here are the notes of today's meeting in irc.freenode.net #blendercoders.
1) Blender 2.70 release candidate
- This week we'll do the 2.70 RC. Build 'ahoy' will be done Wednesday.
- Remember: we now make 'release candidates' complete with splash and correct release number, as if it's a real release. Git 'master' gets copied to an release branch, where bug fixes will be ported to for the final release. That means that within a day after the 'Ahoy' development can continue as usual.
- Brecht van Lommel and Campbell Barton coordinate release and logs.
- Sergey Sharybin will do the git branching & tagging magic.
- Ton Roosendaal will provide a splash, made by the Gooseberry project.
- For now: only total crucial fixes, also no UI label changes (for translations).
- Python Numpy module will only be in Linux/OSX builds. Windows NumPy compiling has issues to tackle still. Martijn Berger will do an attempt to get it solved though.
- Campbell proposes to include the Python 'requests' module (to handle http). This is needed to make the add-on from SketchFab to work, but it's a generic module useful for everyone.
Laters,
-Ton-
8 Comments
including ‘requests’ module would be great !
was just thinking the same thing, I am going to use it really soon.
I'm happy someone is working on windows numpy! I'm building tool that involves some matrix inversion, but I can't use the mathutils tools because my matrices may be larger than 4x4.
hm in the past i used to use numpy as well it worked fine on windows, but in that time i never had heard of blender.
numpy is quite good, and can be faster dan mathematica and some other pro math software, i hope you get it working with blender for your needs, all your guys are doing great things.
Numpy works under windows. The problem is that the devs want to distribute a precompiled version and the recent working precompiled build that is available for py3 uses mkl, which apparentyl has a license incompatible with the gpl.
And indeed, numpy is much much faster than mathematica, as numpy does numerical linear algebra and msthematica is incredibly slow at linear algebra. Numpy adds only a little overhead to the libraries it uses (lapack, blas,atlas) whereas mathematica is one big chunk of overhead.
oh then just ask intel if they would like to support blender, mkl is an intel dependency if i remember well.
I'm not sure I get what you mean. Mkl is intel's, yes. But not every windows user has it installed, so one cannot rely on the lib being there and because of license incompatibility with the gpl you cannot build it statically into a build that you distribute under the gpl. This is not an intel problem. Intel allows you to distribute mkl. The gpl doesn't!
t thought that this license GPL was because of intel, then i am wrong