FMSLogo Code

Implement feature #32; the "X" close button on the undocked commander's title bar is now disabled.

Remove the logic that strips some text off the top of commander history if adding text to the bottom fails. This logic was a vestige from MSWLogo when the control was a simple edit control and Windows 95's edit control had a limit of how much text it could hold. This limit is no longer applicabale, since Windows 95 is not supported and the control is now a wxRichTextCtrl. As a result, the removing-from-top logic was effectively dead code.

Extend the wxWidgets 3.1.0 patch to include a fix for a regression from wxWidgets 2.8.12 that introduced strange glyphs into the logo code editors when some non-printing characters are pressed (F1-F12, media keys, etc.). The bug in wxWidgets is tracked at http://trac.wxwidgets.org/ticket/17598

Fix a false-positive failure that happens when running the test pass where the UI automaticn fails to press the "Halt" button because it cannot find the commander panel. This failed because, after running SPLITSCREEN, the old commander still exists as a window until the application is idle. So if the UI automatation attempts to find a window named "Commander", it may find one that it slated to be destroyed that no longer contains a commander panel.

The fix is to hide the old commander dialog and rename it in a way that computers will see as distinct, but humans won't.

Correct a mistake that Brian Harvey's textbooks are college level. In the introduction, they clearly state that they are intended for high school students (and hobbiest adults). I had assumed because they were textbooks and written by a college professor, that they were for his student.

Fix a regression where the wheel on the mouse didn't scroll the editor. This regression was introduced when I redid the wxWidgets configuration to add only what was necessary, instead of removing what was unnecessary. I had neglected to add mousewheel.

Added an explanation of how the array's mutable nature can lead to aliasing bugs, where modifying an array ends up modifying other references to it.

This confusion was discussed in the Open Discussion forum.

Update the documentation for READWORD to more clearly explain how characters can be "processed" and "preserved" and also remove the incorrect suggestion that a programmer can determine exactly what was entered by using BACKSLASHEDP.

This change was motivated by posting on in the Open Discussion forum.

Speed up the test pass by working around a performance regression when the commander is resized when there is a lot of history. The change is to populate the history with a pre-determined amount of text and create a new test case for when there's a lot of history, which won't be integrated into the history until the bug is fixed.

Update the expected test log to include the results of some new tests which were added that cover POLYSTART, POLYEND, SCROLLX, SCROLLY, DIALOGFILEOPEN, and DIALOGFILESAVE.