2316: Refactoring preferences

[krzywon]

Description

This hopes to combine parts of what was done in #2306 and #2321, specifically related to formalizing the preferences panel and widget class structures.

As asked for in #2316, the following has been added to this PR:

• Settings are staged in the preferences panel until they are accepted and only then is the config updated (using the OK or Apply buttons)
• All staged changes are discarded if they are rejected (using the Close button or X)
• Restore defaults resets the active widget to the default values as read in from the configuration
• Added 'Cancel' and 'Apply' buttons
• Updated the help documentation to include expected behaviors and information on specific

Fixes #2312
Fixes #2313

How Has This Been Tested?

• Unit tests that are in this PR
• From run.py:
  • Open preferences, modify settings, click Cancel, open preferences, and be sure previous changes are gone
  • Open preferences, modify settings, click X, open preferences, and be sure previous changes are gone
  • Open preferences, modify settings, click Ok, open preferences, and be sure previous changes are persistent
  • Open preferences, modify settings, click Apply, click Close and/or X, open preferences, and be sure previous changes are persistent
  • Open preferences, modify settings, click Apply, click Close and/or X, open preferences, click Restore Defaults, and be sure values reset to defaults

Review Checklist (please remove items if they don't apply):

• Code has been reviewed
• Functionality has been tested
• Windows installer (GH artifact) has been tested (installed and worked)
• MacOSX installer (GH artifact) has been tested (installed and worked)
• User documentation is available and complete (if required)
• Developers documentation is available and complete (if required)
• The introduced changes comply with SasView license (BSD 3-Clause)

[lucas-wilkins]

You could use a QFormLayout to make things line up.

[lucas-wilkins]

I've added type checking to config setting because currently the text fields are not validated (will add a config type checker next). This causes some pretty bad issues if you set an int to a string, for example (I tested it).

[lucas-wilkins]

config will now raise a TypeError if you try and give it the wrong type. This would still need to be handled on the gui side (use .validate(name, value) to just test an entry

[krzywon]

Thanks for the info. I'll add Qt validators for ints and floats and apply them to the appropriate inputs.

[gonzalezma]

I think that the default option for Plotting Options/Use full-width ... should be that it is not checked. Otherwise one gets often really ugly plots with the legend appearing just in the middle of the screen. The alternative would be that for each plot the legend size is resized to the length of the longer label, but this will be harder to implement. With this option unchecked we recover the same behaviour as before, while given the user the possibility to modify that whenever the data labels are too large. Also after playing with the options and changing the legend length a couple of times I got an error:
File "C:\Users\gonzalezm\Git\sasview\src\sas\qtgui\MainWindow\GuiManager.py", line 1294, in showPlot
self.filesWidget.displayData(plot, id)
File "C:\Users\gonzalezm\Git\sasview\src\sas\qtgui\MainWindow\DataExplorer.py", line 1087, in displayData
self.plotData([(plot_item, plot_to_show)])
File "C:\Users\gonzalezm\Git\sasview\src\sas\qtgui\MainWindow\DataExplorer.py", line 1156, in plotData
new_plot.plot(plot_set, transform=transform)
File "C:\Users\gonzalezm\Git\sasview\src\sas\qtgui\Plotting\Plotter.py", line 243, in plot
if len(l)> config.FITTING_PLOT_LEGEND_MAX_LINE_LENGTH:
TypeError: '>' not supported between instances of 'int' and 'str'

So probably the plotting options need more testing until finding the best default params.

[gonzalezma]

In the doc, I suppose this is not a % scaling. If so remove percent from
QT Screen Scale Factor: A percent scaling factor ...
And trying to change this factor, I got the following error:
Cannot set set variable 'QT_SCALE_FACTOR', improper type (Cannot coerce <class 'str'> to bool)
Could this be related to the type checking?

[lucas-wilkins]

@gonzalezma

'QT_SCALE_FACTOR', improper type (Cannot coerce <class 'str'> to bool) - Could this be related to the type checking?

Yes, that would be - though indirectly. Not sure why it would say bool, unless something is wired up wrong. It should be able to coerce a string to a float if it is valid (I think)

[lucas-wilkins]

@gonzalezma

Nope.

1. it wont convert strings to floats
2. the error message for this says bool, oops

[krzywon]

I've added validators to the float and integer inputs and ensured the values are coerced before hitting the config

[butlerpd]

moving legend location to be put into a new PR. Lucas to do proper review and hopefully approve