Add display marker position and origin feature to this method:

https://github.com/OrtnerMichael/magPyLib/blob/e0951f729e8a5155ab3f9827b7d196898327fd7c/_lib/classes/collection.py#L190

pm = magpy.magnet.Cylinder(mag=[0,0,1000],dim=[5,5])
for x in linspace(0,2.5,5):
for z in linspace(0,2.5,5):
print(pm.getB([x,0,z]))

Is there a better way for initialization of modules to achieve tool tips as desired?
This should work nicely for Spyder, Jupyter and Pycharm

Current structure:
.
├── current
│   └── __init__.py
├── __init__.py
├── _lib
│   ├── classes
│   │   ├── base.py
│   │   ├── collection.py
│   │   ├── currents.py
│   │   └── magnets.py
│   ├── fields
│   │   ├── Current_CircularLoop.py
│   │   ├── Current_Line.py
│   │   ├── Moment_Dipole.py
│   │   ├── PM_Cube.py
│   │   ├── PM_Cylinder.py
│   │   └── PM_Sphere.py
│   ├── mathLibPrivate.py
│   └── mathLibPublic.py
├── magnet
│   └── __init__.py
├── math
│   └── __init__.py
├── README.md
└── RotateExample.py

Some machines' Spyder build cannot import the "source" package albeit the installation shows it is in place.

see title

• Maybe use a line coming out of the sources to indicate the magnetization
• Arrows for line currents ?
• Alternative ideas?

https://github.com/OrtnerMichael/magPyLib/blob/e0951f729e8a5155ab3f9827b7d196898327fd7c/_lib/classes/collection.py#L190

• Update once V1.0 is ready

• maybe include graphics in docstrings on webpage

Michael: this can be part of the documentation pdf
Lucas: It may be possible on the webpage Doku

• Possible autodoc tools for Docstrig integration and more:
  • Sphinx - Most Popular
  • Doxygen - Less friendly, more exotic exports (raw LaTeX)
• Possible Hosting sites for autodocs (update automatically):
  • https://readthedocs.org/ - Seems the simpler
  • https://pages.github.com/ - Most Popular though most general
  • https://www.gitbook.com - Most professional looking, free open source

• Docstrings -> HTML

• Possible autodoc tools for Docstrig integration and more:

  • [Sphinx][1] - Most Popular
  • [Doxygen][2] - Less friendly, more exotic exports (raw LaTeX)
    [1]: https://www.sphinx-doc.org/en/master/index.html
    [2]: http://www.doxygen.nl/
    [3]: https://www.oreilly.com/ideas/the-eight-rules-of-good-documentation

• Possible Hosting sites for autodocs (update automatically):

  • https://readthedocs.org/ - Seems the simpler
  • https://pages.github.com/ - Most Popular though most general
  • https://www.gitbook.com - Most professional looking, free open source

• documentation on webpage

• Maybe combine auto-generated HTML with additional features like graphics

• detailed pdf documentation

  • 1. Programming Class descriptions
  • 2. Mathematical package doku how is it structured (Coordinate Systems, Translations + Rotations, ...)
  • 3. Phys doku: how are fields calculated, collect the papers where formulas can be found and upload or generate a document with all calcuations
  • 4. Physics Plausibility checks:
    • a. compare to FEM
    • b. check precision/convergence of diametral cylinder class
    • c. characterize demagnetization

Perla's code works fine with sequential input, but on Windows it hangs.

https://docs.python.org/3/library/multiprocessing.html#examples

As per given, Windows handles creating of different processes differently.

When users make code they want to use with multiprocessing, if they are on Linux:
Linux code is valid:

import time
from magpylib import source

def getPos():
        return [2,2,2]

a = source.magnet.Box(mag=(2,3,5),dim=(2,2,2))
start = time.time()
listOfPos=[getPos() for a in range(5000)] # 5000 positions
result = a.getBsweep(listOfPos,multiprocessing=True)
end = time.time()
print(end - start)
print(result)

If they are on Windows however, the following guard is necessary to avoid lock up:

import time
from magpylib import source

def getPos():
        return [2,2,2]
def main():
        a = source.magnet.Box(mag=(2,3,5),dim=(2,2,2))
        start = time.time()
        listOfPos=[getPos() for a in range(5000)] # 5000 positions
        result = a.getBsweep(listOfPos,multiprocessing=True)
        end = time.time()
        print(end - start)
        print(result)

from multiprocessing import freeze_support # windows support
if __name__ == '__main__':  #################### These two define 
    freeze_support() ########################### the Windows OS
    main()

• Change name of Cube to Box
• Fix all dependencies e.g. docstrings etc

put all cool stuff there

"FieldSampler"

this class is inherited by RCS and by collection!!
this class initiates the getB, getH, getF methods and provides unified docstrings for all

Analyze initialization for must-have kwargs like "mag=None", "dim = None"

• Propose alternatives to "None" after user-friendliness, as users should immediately realize that they have to set those values.

https://github.com/OrtnerMichael/magPyLib/blob/e0951f729e8a5155ab3f9827b7d196898327fd7c/_lib/classes/magnets.py#L99

• docstring ctrl+c ctrl+v problem

Add fundamental tests for the lines:

pm = magpy.source.magnet.Box(mag=[0,0,1000],dim=[0.5,0.1,1],pos=[.25,.55,-1111])
print(pm.getB([.5,.5,5]))

this returns NAN NAN NAN!!!!!

• Multi getB restructure ( #69 )
• At first make this a new method, then later integrate with getB functions as option
• input vector of positions
• introduce use of parallel computing to calculate points in parallel
• return vector of fields

• Proper checking of inputs and input formats

Make sure invalid inputs will be notified
Make toolbox foolproof

• Check performance reduction of input type checking - input type checking should not decrease performance

• Consider feasibility of Debug mode with asserts
  • proper error handling, braking of program and return msg
  • should negate performance impact as it disables checks outside of debug mode

• webpage
  • check git hub pages possibility
  • wordpress as alternative
  • NO ADVERTISEMENT
• Git hub readme
• Read the docs - use Markdown
  • check out alternatives.

The idea is to introduce a new class type that mimics a sensor with position and orientation that can determine the field from existing source objects.
This has the advantage that sensor positions and especially rotations must not be implemented b the user.

magpy.sources.magnet
- logic structure, better for beginners to ensure what are the sources
- longer code is annoying

• 3d plot z-order problems in display system, specifically when bounding boxes are overlapping

• in current version of Spyder tool tips sometimes do not display.

  • start in project mode
  • wait till Spyder has caught up

rename "CoR" as "anchor"

• replace name in all methods and classes
• replace name in all doc strings
• any other issues?

• REVIEW GUIDELINE

what has to be done when a completely new source class is added
point out all positions in the code where one must do something :)

• docstrings format to numpy native

• Fix outdated examples in docstrings

• Check alternatives for placement of docstrings (i.e. keep them in an external file)

Lucas: not a good practice as far as guidelines go,
while feasible in external .txt file + making a tool for release

• getB - how to avoid multiple times the same docstring (in each case there is only one line different)

Lucas: solution depends entirely on tool being used to analyze docstring placement,
ctrl+c ctrl+v seems to be the universal solution so far.
Could still explore more of class inheritance and overriding for this, though it
would take major refactoring of existing code.

• REVIEW INSTALLATION TIPS

How to best distribute the library?

• download from GITHUB or webpage and use as separate "folder" in each program
  • no so practical
  • no problems with backwards compatibility
• integration in pip
  • backwards compatibility problems?
• integration with Python package index (PyPI)
  • easy for all users
  • keeping it running over the years?
• add to conda cloud

• Give an option to display a motion.

  • Is it worth the effort?

• Display current direction and magnetization direction

  • arrows on current lines
  • put thin axes "through" magnets that indicate the magnetization vector

• Introduce Markers of positions

  • e.g. to indicate anchor points
  • by default introduce marker at center ?

• fix plt.cm package use

  • is this just some outdated code?

👍

• dipole sizes may vary (relative definition to SYSSIZE)
  • If dipoles get initialized at expanding coordinates, their size will vary.
• Coose better color function and color assignement

Currently shows up as nothing on display. Add this to the collection list.

https://github.com/OrtnerMichael/magPyLib/blob/e0951f729e8a5155ab3f9827b7d196898327fd7c/_lib/classes/collection.py#L190

Show up as an arrow in the 3D plot when using Collection.displaySystem()

1. Test all 8 quadrants for each source (INSIDE AND OUTSIDE of magnet)
2. UNDERSTAND WHAT WE WANT TO TEST: SPECIAL CASES (LINES, CORNERS, SURFACES, ...)

pm = magpy.source.magnet.Box(mag=[0,0,1000],dim=[0.5,0.1,1],pos=[.25,.55,-1111])
print(pm.getB([.5,.5,5]))
this returns NAN NAN NAN!!!!!

After testing structure for all the getBs are set up, just define the quadrants to reinforce robustness.

Figure out Github

figure out the license structure

• Contributor's License details
• Repository License

This should removes one of the sources from the collection list.
Useful for terminal use.

• Automated unit tests within test suites for development
  • Very brief "Black/White Box" (check if input -> output always) testing scripts for each function
  • Will allow us to make code modification experiments freely (while after performance, etc) as it notifies inconsistencies in behavior
    • Could also be useful in implementing new functions since we know what we want
    • Counts as documentation! Usually only have to write once, add more when needed
  • Automated Tools for open projects run the tests, hold onto clearly broken updates + let us know
    • Travis CI

mag = (0,0,0) rather than mag = (0.0,0.0,0.0)

0-point shift on cube is limited to 1e-6.
WHY ?

simple static webpage only with links to GitHub and Doku

NO COMMERCIALS !!!!

At class init all args are kwargs for class methods they are not.

Does it make sense to make all args kwargs?

• better for understanding
• worse for shortening and simplifying codes