Bugfix #48
Bugfix #48
Conversation
Previously, the Bluestein optimization were not regarded in the complexity model, effectively causing some trouble. Also reduce code duplicates in forward and backward by deriving the backward from the forward routine.
Previously, an error was thrown when a tap position was specified that exceeded the size of the register. As the tap configuration directly corresponds to the generating polynomial -- and this always has a x^k factor with k being the order of the polynomial (size of the register) -- it is sensible to also accept the full notation of the polynomial.
- Rename .regSize -> .size - Rename .regTaps -> .taps - Rename .resetState -> .start - Add .period property reporting the period of the sequence generated by the register configuration - Remove `length`option from initialization as the intended feature behind this option is currently broken and was previously not documented to the user. Instead a description on how it could be introduced later-on was added as a comment at the end of __init__() - Remove default value for the `start` option to __init__() to make the actual register configuration explicit
Testing the previous interface extension commits resulted in false-positives for these (and deducted) classes as a pip-installed version shadowed the version-under-test. This is now resolved.
- Add __init__(minType=<type>) option, applicable to all classes. This option allows the user to widen the input datatype to mitigate range overflows. - Introduce internal ._prepareInputArray() method to handle all input array related stuff at one place for ._forward(), ._backward() and their corresponding cython-type methods. This method now also includes the dimension checks and cleans up code clarity of .forward() and .backward() - Introduce TRANSFORM struct being a container to keep track of input-, intermediate- and output array data types and shapes.
|
So commit 124cf96 "cleans up code clarity"? Does this introduce some much needed chaos? :3 |
|
Yeah. It's true. Totally clear to me. So sad. |
- Improved consistency of additions to **options during __init__: You should never ADD an options to the **options dict to avoid interference with the userspace. Instead, copy the dict in options and modify the copy. This is now enforced. - Rename 'forceInputAlignment' to 'forceContiguousInput' - Rename argument in estimateRuntime - Rename 'useFortranStyle' to 'fortranStyle' - Rename 'minType' to 'minFusedType' internally as the external parameter is a dtype and the internal a fused type number (indexed) - Modified parameter passing in base class instantiation of transposed or conjugated classes to apply them by-dictionary, which gets extracted with Matrix._getProperties. This is more future-safe - Modified class-dependent setters of options `forceInputAlignment`, `bypassAutoArray`, `widenInputDatatype` and `fortranStyle` to use direct assignments instead of utilizing the parameter passing.
Believe me. It's true.
There was a problem hiding this comment.
traceback (most recent call last): File "setup.py", line 376, in <module> ext_modules=lazyCythonize(extensions) File "/usr/lib/python3.7/site-packages/setuptools/__init__.py", line 140, in setup return distutils.core.setup(**attrs) File "/usr/lib/python3.7/distutils/core.py", line 148, in setup dist.run_commands() File "/usr/lib/python3.7/distutils/dist.py", line 966, in run_commands self.run_command(cmd) File "/usr/lib/python3.7/distutils/dist.py", line 984, in run_command cmd_obj.ensure_finalized() File "/usr/lib/python3.7/distutils/cmd.py", line 107, in ensure_finalized self.finalize_options() File "/usr/lib/python3.7/site-packages/setuptools/command/build_ext.py", line 133, in finalize_options _build_ext.finalize_options(self) File "/usr/lib/python3.7/site-packages/Cython/Distutils/old_build_ext.py", line 167, in finalize_options _build_ext.build_ext.finalize_options(self) File "/usr/lib/python3.7/distutils/command/build_ext.py", line 140, in finalize_options ('plat_name', 'plat_name'), File "/usr/lib/python3.7/distutils/cmd.py", line 287, in set_undefined_options src_cmd_obj.ensure_finalized() File "/usr/lib/python3.7/distutils/cmd.py", line 107, in ensure_finalized self.finalize_options() File "/usr/lib/python3.7/distutils/command/build.py", line 105, in finalize_options if self.distribution.ext_modules: File "setup.py", line 150, in __len__ return len(self.c_list()) File "setup.py", line 138, in c_list self._list = self.callback() File "setup.py", line 204, in extensions compiler_directives=cythonDirectives File "/usr/lib/python3.7/site-packages/Cython/Build/Dependencies.py", line 1086, in cythonize cythonize_one(*args) File "/usr/lib/python3.7/site-packages/Cython/Build/Dependencies.py", line 1192, in cythonize_one result = compile_single(pyx_file, options, full_module_name=full_module_name) File "/usr/lib/python3.7/site-packages/Cython/Compiler/Main.py", line 725, in compile_single return run_pipeline(source, options, full_module_name) File "/usr/lib/python3.7/site-packages/Cython/Compiler/Main.py", line 513, in run_pipeline err, enddata = Pipeline.run_pipeline(pipeline, source) File "/usr/lib/python3.7/site-packages/Cython/Compiler/Pipeline.py", line 355, in run_pipeline data = run(phase, data) File "/usr/lib/python3.7/site-packages/Cython/Compiler/Pipeline.py", line 335, in run return phase(data) File "/usr/lib/python3.7/site-packages/Cython/Compiler/Pipeline.py", line 52, in generate_pyx_code_stage module_node.process_implementation(options, result) File "/usr/lib/python3.7/site-packages/Cython/Compiler/ModuleNode.py", line 143, in process_implementation self.generate_c_code(env, options, result) File "/usr/lib/python3.7/site-packages/Cython/Compiler/ModuleNode.py", line 379, in generate_c_code self.body.generate_function_definitions(env, code) File "/usr/lib/python3.7/site-packages/Cython/Compiler/Nodes.py", line 442, in generate_function_definitions stat.generate_function_definitions(env, code) File "/usr/lib/python3.7/site-packages/Cython/Compiler/Nodes.py", line 442, in generate_function_definitions stat.generate_function_definitions(env, code) File "/usr/lib/python3.7/site-packages/Cython/Compiler/Nodes.py", line 4839, in generate_function_definitions self.body.generate_function_definitions(self.scope, code) File "/usr/lib/python3.7/site-packages/Cython/Compiler/Nodes.py", line 442, in generate_function_definitions stat.generate_function_definitions(env, code) File "/usr/lib/python3.7/site-packages/Cython/Compiler/Nodes.py", line 1942, in generate_function_definitions self.generate_argument_parsing_code(env, code) File "/usr/lib/python3.7/site-packages/Cython/Compiler/Nodes.py", line 2626, in generate_argument_parsing_code self.type.opt_arg_cname(declarator.name))) File "/usr/lib/python3.7/site-packages/Cython/Compiler/PyrexTypes.py", line 2957, in opt_arg_cname return self.op_arg_struct.base_type.scope.lookup(arg_name).cname AttributeError: 'NoneType' object has no attribute 'cname'
Unfortunately I get this error with cython 0.29 and CPython 3.7.1. after removing some obvious and minor typos in Matrix.pyx and MLCiruclant.pyx :-/
|
I shouldn't have cherry-picked the docu stuff from the int16 branch where it was pushed accidentially. Could have been easier just to accept it there, when this branch was still working ¯_(ツ)_/¯ |
|
Well now that shit has hit the fan, how do we proceed? :3
…On Fri, 9 Nov 2018 at 10:05, Christoph Wagner ***@***.***> wrote:
I shouldn't have cherry-picked the docu stuff from the int16 branch where
it was pushed accidentially. Could have been easier just to accept it
there, when this branch was still working ¯_(ツ)_/¯
—
You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub
<#48 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABDZRuftyHfSqSmyLd3kq9ScwJdYlTLiks5utUVmgaJpZM4YTBBg>
.
|
|
I do what I should hae done in the first place: Merge in int16 and do both pull requests in one. Currently having IT problems. Hopefully the VM starts after lunch or there is some shit that will hit the fan at hp support ;) |
|
Alright. Let's see how this plays out. :) Good luck with the HP!
…On Fri, 9 Nov 2018 at 11:22, Christoph Wagner ***@***.***> wrote:
I do what I should hae done directly. Merge in int16 and do both pull
requests at one. Currently having IT problems. Hopefully the VM starts
after lunch or there is some shit that will hit the fan at hp support ;)
—
You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub
<#48 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/ABDZRgfSIJekMhTmME2t8Xc_EmxBjlsdks5utVdogaJpZM4YTBBg>
.
|
|
Thanks, should be fine now. Happy approving! :) |
Some bug fixes and extensions. Resolves issues #37, #39, #40, #41, #42, #44
Previous pull request was withdrawn.