taichi-dev / taichi Goto Github PK

They sometimes return 1...

I'm getting tired testing the environment on all the 3 platforms and I think it's better to have CIs for Taichi. Please let me know it you have any suggestion on this topic or want to help realize it. Thank you. It looks that Travis CI for Linux and Mac OSX and AppVeyor for Windows are suitable choices.

On MacOSX(10.11.6), some errors will happen when execute nbody_3d.py.
I fixed them in a not clever way, but it works for me.

Traceback (most recent call last):
 File "nbody_3d.py", line 1, in <module>
   from taichi.dynamics.nbody import NBody
 File "/Users/YJ-work/Desktop/taichi_root/taichi/python/taichi/dynamics/nbody.py", line 6, in <module>
   from taichi.tools.video import VideoManager
 File "/Users/YJ-work/Desktop/taichi_root/taichi/python/taichi/tools/video.py", line 1, in <module>
   from taichi import ndarray_to_array2d
ImportError: cannot import name ndarray_to_array2d

I replaced from taichi import ndarray_to_array2d with from taichi.misc.util import ndarray_to_array2d in taichi.tools.video.

[Errno 2] No such file or directory: '/Users/YJ-work/Desktop/taichi_root//taichi_outputs///task-2017-01-10-21-51-00-r07015'
Simulation time: 0.0
nbody.cpp[201]: node_end = 297117
Time: 0.0983819961548
Segmentation fault: 11

I replaced self.directory = taichi.get_output_path(get_unique_task_id()) with a full path self.directory = '/Users/YJ-work/Desktop/taichi_root/taichi/output/' in taichi.dynamics.nbody

Traceback (most recent call last):
  File "nbody_3d.py", line 6, in <module>
    nbody.step(0.01)
  File "/Users/YJ-work/Desktop/taichi_root/taichi/python/taichi/dynamics/nbody.py", line 49, in step
    img = LDRDisplay(exposure=1, adaptive_exposure=False).process(img)
NameError: global name 'LDRDisplay' is not defined

I added from taichi.visual.post_process import LDRDisplay in taichi.dynamics.nbody.

Traceback (most recent call last):
  File "nbody_3d.py", line 6, in <module>
    nbody.step(0.01)
  File "/Users/YJ-work/Desktop/taichi_root/taichi/python/taichi/dynamics/nbody.py", line 53, in step
    self.video_manager.write_frame(img)
  File "/Users/YJ-work/Desktop/taichi_root/taichi/python/taichi/tools/video.py", line 18, in write_frame
    ndarray_to_array2d(img).write(self.directory + '/%05d.png' % self.frame_counter)
AttributeError: 'NoneType' object has no attribute 'write'

I commented out self.video_manager.write_frame(img) in taichi.dynamics.nbody.

Maybe there is a clever way to fix this problem.

BTW, I got this output (one of those frames)

Is it correct?

Keep getting this issue when installing it with make -j 2

fatal error: taichi/dynamics/poisson_solver2D.h: No such file or directory
#include <taichi/dynamics/poisson_solver2D.h>
^
compilation terminated.
make[2]: *** [CMakeFiles/taichi_core.dir/src/simulation3d/poisson_solver2d.cpp.o] Error 1
make[2]: *** Waiting for unfinished jobs....
make[1]: *** [CMakeFiles/taichi_core.dir/all] Error 2
make: *** [all] Error 2

I followed the 16.04 instructions to install on a new Ubuntu 16.10 VM but the library ended up being linked with -lpython3.5m and would not work.

To link with python 2.7 I had to use this cmake command:

cmake \
  -DPYTHON_INCLUDE_DIR:PATH=/usr/include/python2.7 \
  -DPYTHON_EXECUTABLE:FILEPATH=/usr/bin/python2.7 \
  -DPYTHON_LIBRARY:FILEPATH=/usr/lib/x86_64-linux-gnu/libpython2.7.so \
  ..

Taichi won't import:

/home/janvorac/wheel/packages/taichi/taichi/python/taichi/core/load_core.py in <module>()
     29         sys.path.append(bin_dir)
     30         shutil.copy('libtaichi_core.so', 'taichi_core.so')
---> 31         import taichi_core as tc_core
     32 
     33         os.chdir(tmp_cwd)

ImportError: ./taichi_core.so: undefined symbol: __cpu_model

Installed using your wiki guide on XUbuntu 16.04.02. I am using Python 2.7.12 in virtual environment downloaded via mkvirtualenv (not compiled on my PC).
My gcc/g++ version is 5.4.0, cmake version is 3.5.1. Any help would be welcome!

I'm using Xcode building the project, but caught these errors:

1. boost header including error:

fatal error: 'boost/foreach.hpp' file not found

Fixed by adding the boost including path to "search paths".

2. a link error:

clang: error: no such file or directory: '/Users/Iterator/homebrew/Cellar/python/2.7.8_1/Frameworks/Python.framework/Versions/2.7/lib/libpython2.7.dylib'

It seems that I should change "/Users/Iterator/homebrew" to my own path(in Build Settings--Other Linker Flags).
For me, it is "/usr/local/Cellar/python/2.7.10_2/Frameworks/Python.framework/Versions/2.7/lib/libpython2.7.dylib"

They should be objects...Mesh should be used for triangle meshes.

It would be good to have an additional browser viewer for rendering/simulation process visualization. Python itself has good HTTP server support and it will help in case people use Taichi for some task via ssh and do not want to enable X11 forwarding.

After I built libtaichi_core.dylib and run the test code, I have this error

>>> import taichi as tc
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/YJ-work/Desktop/taichi_root/taichi/python/taichi/__init__.py", line 1, in <module>
    from taichi.core import tc_core as core
  File "/Users/YJ-work/Desktop/taichi_root/taichi/python/taichi/core/__init__.py", line 2, in <module>
    from load_core import tc_core
  File "/Users/YJ-work/Desktop/taichi_root/taichi/python/taichi/core/load_core.py", line 18, in <module>
    import taichi_core as tc_core
ImportError: dlopen(./taichi_core.so, 2): Library not loaded: libboost_system.dylib
  Referenced from: /Users/YJ-work/Desktop/taichi_root/taichi/build/taichi_core.so
  Reason: unsafe use of relative rpath libboost_system.dylib in ./taichi_core.so with restricted binary

I googled and found the problem is in libtaichi_core.dylib:

aibm8:build YJ-work$ otool -L libtaichi_core.dylib 
libtaichi_core.dylib:
	/Users/YJ-work/Desktop/taichi_root/taichi/build/libtaichi_core.dylib (compatibility version 0.0.0, current version 0.0.0)
	/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1226.10.1)
	/opt/local/lib/libembree.2.dylib (compatibility version 2.0.0, current version 2.13.0)
	/System/Library/Frameworks/Python.framework/Versions/2.7/Python (compatibility version 2.7.0, current version 2.7.10)
	/opt/local/lib/libboost_system-mt.dylib (compatibility version 0.0.0, current version 0.0.0)
	/opt/local/lib/libboost_python-mt.dylib (compatibility version 0.0.0, current version 0.0.0)
	libboost_system.dylib (compatibility version 0.0.0, current version 0.0.0)
	libboost_python.dylib (compatibility version 0.0.0, current version 0.0.0)
	/usr/lib/libc++.1.dylib (compatibility version 1.0.0, current version 120.1.0)

And I changed the path libboost_system.dylib, libboost_python.dylib to:

/opt/local/lib/libboost_system.dylib (compatibility version 0.0.0, current version 0.0.0)
/opt/local/lib/libboost_python.dylib (compatibility version 0.0.0, current version 0.0.0)

then the error is gone.
The problem is: every time I rebuild libtaichi_core.dylib, I have to repeat the above things again.
Is there any way to solve this problem?

/opt/local/lib is where I put libboost_system.dylib and libboost_python.dylib.

I plan to first have the documentation on the Python part, so that people can use it. The doc for C++ part will be put lower priority.

JSON + Typed array or BSON?

It's just used in python for Guassian blur.

I'm sorry that this taichi hasn't been updated for a while. This is because I used it for my research and some part of code cannot be open source before the paper is published. However, I'll merge research-unrelated part into this repos.

We need one that supports C++11, smart_ptr (avoiding duplication)...

A good choice seems to be

https://github.com/USCiLab/cereal

Possibly replace the dependency with Kivy or PIL?

The 13th line of test.py
show_image('img', img)
doesn't work. It never response.

When I comment out show_image('img', img), the program ends correctly.

Maybe there is something wrong with show_image?

Please add info to "Getting Started" how compile on windows.
Thanks.

UUID is a well-known identifier standard with a format like 123e4567-e89b-12d3-a456-426655440000. This function does not generate UUIDs.

Hello. Will support this iterator for go language? I also have own path tracing, but for different purpose, so have no much functions. I know fogleman but I seen your iterator more advanced.

It turns out that sometimes users have to specify a lot of parameters for a simple function call.

A cheaper way is to just render multiple frames and then somehow take an average.

~/taichi/taichi/python/examples/simulation$ python nbody_3d.py

[Errno 2] No such file or directory: '../output/frames/task-2017-01-05-20-58-23-r03610/'
Simulation time: 0.0
nbody.cpp[201]: node_end = 297117
Time: 0.139370918274
Segmentation fault (core dumped)

I think the error is nbody.py where this line:
self.directory = '../output/frames/' + get_uuid() + '/'

Maybe should be:
self.directory = tc.get_root_directory() + 'output/frames/' + get_uuid() + '/'

added this:
import taichi as tc

but then get this error:

python nbody_3d.py 
[Errno 2] No such file or directory: '/home/me/taichi/output/frames/task-2017-01-05-21-19-15-r07527/'
Simulation time: 0.0
nbody.cpp[201]: node_end = 297117
Time: 0.115436077118
Segmentation fault (core dumped)

as if it cannot os.mkdir(self.directory)

Getting tired dealing with templates and if constexpr would be a great help. We should make the switch when all platforms have C++17 ready.

Hi, I built libtaichi_core.dylib successfully, following the Mac OS X part of Wiki.
But when I run
$python2.7
>>> import taichi as tc
I got

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ImportError: No module named taichi

What should I do?

include_directories(${BOOST_INCLUDE_DIRS})
中Boost是全大写，但是
find_package(Boost COMPONENTS system python REQUIRED)
中Boost是区分大小写，导致osx下无法寻找到正确的路径

It would be great if users can install Taichi just by

pip install taichi

Actually I had a try days ago but it's interesting that I can only package the *.py files in taichi/python/, even if I added the rest to MANIFEST.in.

If you have any idea on what's going on, please let me know.
Thanks!

They all seem to fail with the same assertion:

$ python bubbles.py
asset_manager.h@(Ln 20): Assertion Failed. [Asset has been expired]

My taichi is compiled against Python 2.7.13, boost 1.6.13, embree 2.13.0, and tbb 4.4-20161128.

The examples that do work are:

• geometry
• scoping
• trees

Maybe taichi could be split into several modules:

1. (Module A) C++ graphics rendering engine: Given a 3D scene in a standard data format (e.g. the one in taichi/blender/exporter/__init__.py), the engine calculates a rasterised image. This rendering engine could be used interchangeably with other rendering engines.

2. (Module B) Python library to procedurally generate a 3D scene. The output is given in a standard data format, and can be fed into any graphics rendering engine.

3. (Module B) C++ physics simulation: given some parameters, calculates a series of scenes. This part could be integrated into Module B: this can be seen as a specific way to generate 3D scenes.

4. (Module C) A seperate Python library to handle realtime preview and program hot reloading, for any Python program that generates a series of images. We don't need to have lovecat integrated into Module C.

5. (Module D) A set of demos that is built upon Module B, Module C and lovecat. Binary asset files could go into Module D.

There are good ones but supports Linux only:
https://github.com/bombela/backward-cpp

We need a cross-platform and non-GPL licensed (header-only if possible) library.