ENH: Add basic Windows support #7
Conversation
Center the cutoff frequency Magnitude and phase plots in normal Bode plot layout on top of each other with same freq axis instead of side by side also PEP8 fixes
I was getting something like "1.200000000002" in my graph also PEP8 fixes to all examples
This assumes a particular directory structure (but then again, so does ngspice's spinit file). It changes working directory to load the DLL and dependencies without problems. This might not be optimal, but it works.
Add grids, sharex Change x margins to 0 for MPL 2.0 Updated image with all changes Use Pythonic ac() command instead of cmd()
This was not working intermittently because ngspyce.vectors(['Vcos', 'Vsin', 'time']) is a dictionary and the outputs aren't in random order.
Includes backwards incompatible change: vectorNames -> vector_names "Function names should be lowercase, with words separated by underscores as necessary to improve readability. mixedCase is allowed only in contexts where that's already the prevailing style (e.g. threading.py), to retain backwards compatibility."
"Avoid raising a generic Exception, to catch it, you'll have to catch all other more specific exceptions that subclass it." Also reword error message for ac sweep validation
Add tests for ac(), cmd(), model_parameters(), linear_sweep() to improve test coverage
Make sure that spinit and default *.cm files are being loaded. These include spice2poly which is required for most vendor op-amp models, etc.
|
Hmm I guess this is including all changes I've made to my master branch. Not sure how best to handle that. |
Change examples to use it instead of cmd() Add a test for it
# Conflicts: # examples/lowpass/lowpass.py # examples/npn/npn.py # examples/quadrature_oscillator/quadosc.py
|
Looks good. Do the tests pass on Linux? |
|
I'll try to compile ngspice for Linux, never tried that before |
|
Hmm, I cloned the ngspice git repo ( |
|
Ok I got ngspice-26.tar.gz instead and the instructions worked this time. |
|
If I edit
In my testing, many spice models require the spice2poly.cm to be enabled, and they weren't working in Windows until I figured out the paths, etc. so I wrote test for them. |
|
Don't worry, I'll test it in Linux.
…On Feb 11, 2017 3:52 PM, "endolith" ***@***.***> wrote:
If I edit /usr/local/lib/python3.5/dist-packages/ngspyce/ngspyce.py and
change the line to spice = CDLL('/usr/local/lib/libngspice.so') then
ngspyce works, but the tests for the .cm files don't work. So I guess
they're not enabled by default? So
- if those are optional to ngspyce, change the tests into warnings
- if those are required for ngspyce, change the build instructions to
enable them
In my testing, many spice models require the spice2poly.cm to be enabled.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#7 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AAY7KwmtT0fsMo3Cp6D3_m4YE98VbORHks5rbgNfgaJpZM4L3ACm>
.
|
XSpice tests: skip if not compiled, only test default libraries
|
Thank you! |
|
So are there things you'd be ok with me merging directly, or should I always file pull requests? Unfinished stuff is in https://github.com/endolith/ngspyce/tree/experimenting |
|
I'd prefer to look at any changes that might break something, and at least run the tests on Linux. By the way, what are you using ngspyce for? |
|
Ok I'll just file PRs. :) I've only done some experiments so far, but I'd like to automate some circuit design stuff, like optimizing resistor values to maximize THD+noise for a circuit, etc.
|
This assumes a particular directory structure (but then again, so does
ngspice's spinit file).
It changes working directory to load the DLL and dependencies without
problems. This might not be optimal, but it works.
I tried to add some guessing of 32-bit and 64-bit, and in case system drive is not C:\ but I didn't test anything except 64-bit C:\
Fixes #5