4th release candidate for version 0.1

Sadly, a lot of smaller bugs still had / have to be squashed:

• major rework of the import of filter modules and classes, giving more flexibility (and less errors) in the organization of filter modules. The import and filter tree building process now also fails more graciously with buggy filter modules.
• new default filter class 'Manual' for the manual entry of P/Z and coefficients. FIR / IIR can now be selected in the P/Z and the coefficient entry widgets
• make docutils an optional module
• fixed wrong indication of spec violations in filter_info.py
• lots of code and comment clean-up

3rd release candidate for version 0.1

• fixed some incompatible changes pyQt4 -> pyQt5 in QFileDialog and in hdl_specs.py
• fixed some bugs in file_info.py leading to false spec violations. This does not fix #33 yet.
• made layout a little more compact for small screens by moving the reset button of weight specs and amplitude unit combo box

2nd release candidate for version 0.1

• now compatible with pyQt4 and pyQt5 (#26)
• fixed regression error where dynamic filter import error remained permanent
• fixed a py2/py3 error in select_filter.py (#42)
• implemented display of settling error in plot_impz.py (#29)
• implemented sosfilt for improved impulse response of high-order systems

1st release candidate for version 0.1

Lots of renaming and refactoring and finally! #8 is fixed - the app no longer crashes upon exit.

Sixteenth beta for initial release

• The subwidgets input_amp_specs and input_weight_specs now round the displayed numbers to max. 3 digits. When clicking into the text field, the full precision is displayed
• "Changed specs" only appears when specs have been edited
• Dynamic signal-slot connections have been replaced by eventFilter which does not produce memory allocation faults
• Several small corrections in plot_3d
• In the tab for the impulse response, sine / rect and sawtooth stimuli can be selected and plotted as well
• The UI now has a movable splitter between input and plot widgets. Please give feedback whether this is useful or not.
• Minimum filter order is now be selected using a checkbox instead of a (single) radio button
• All modules can now be run using python -m pyfda..
• Several small bug fixes, code clean up

Fifteenth beta for initial release

• move all icons into a QRC to avoid problems with missing icons on some systems
• various small improvements in packaging (several tries needed with pip, bumping up version every time ...)
• package is now available in wheel format as well

Tenth beta for initial release

This is the first release that has been uploaded to PyPI.

• Some fixes to allow clean upload to PyPI (and more important: download)
• This is also the reason for "beta 10" : PyPI forced to bump up the version again and again ...
• Some more fixes to files export and import

Sixth beta for initial release

• fixed several inconsistencies in setup.py leading to missing icons
• fixed several bugs in input_files.py that mostly affected saving and loading files under Linux
• fixed muddling up of initial corner frequency / amplitude pairs
• adapt logging to changed module structure

Fifth beta for initial release

• changed license to MIT, added Open Iconic icons to license information
• fix setup.py install bug
• improved scaling of 3D plot
• suppress errors in SOS - calculations of complex-valued systems
• make calculation of H_mag more robust against pathological cases like integrator system
• several small bug fixes and refactoring

4th beta for initial release

• enable 'sos' as output format for IIR filters when scipy > 0.16 is installed
• new routines fil_convert and fil_save allowing to calculate filters in more than one output format (e.g. 'ba' and 'zpk') without needing to convert between both formats - important for ma - filter
• start making API for filter modules more transparent
• fix a bug where frequency range sometimes was wrong after zooming out full
• make calculation of phase more robust against inf or nan in H - unwrapped phase could not be calculated at all in this case
• show 'a' column in input_coeffs when manually designed filter (e.g. from input_pz) is IIR