Editor: programmatic widget creation doesn't like focusOnLoad

[Douglas Hays]

The focusOnLoad:true boolean doesn't work well with programmatic Editor creation. It works on Firefox, but fails with IE and WebKit?.

[Douglas Hays]

simple testcase

[Douglas Hays]

Needs review by jaredj

[Douglas Hays]

Related problem: It's common for Editor widgets to be created programmatically and then followed by a call to focus() (InlineEditBox/Dialog?). Even if the focus() call is in a setTimeout, the Editor may not have finished loading yet. Editor's focus() method needs to check for isLoaded and possibly queue the focus.

[Douglas Hays]

Another related issue: Editor's focus method sometimes does a placeCursorAtStart (non-IE). But related to the original problem in this ticket, the call to placeCursorAtStart needs to happen for IE also, and in addition, it should happen anytime setValue is called since this effectively resets the caret position.

[Douglas Hays]

The 2nd problem above can be seen by running test_InlineEditBox.html, locate the Editor test near the bottom and click the text. The input caret is not seen (tested Chrome 4, FF 3, and IE 7.
the last problem can be seen by running test_InlineEditBox.html with Chrome. Click the text and an Editor widget is created. Click cancel, and then click the text again. This time, an Editor widget is NOT created but just setValue is called on an existing Editor widget and then focus() but placeCursorAtStart should have been called but wasn't.

[Douglas Hays]

fix Editor focus issues during load

[Douglas Hays]

(In [21291]) Fixes #10678 !strict. Changed focus() to be deferrable during Editor initialization. After the Editor value is changed, focusing will cause the cursor to go to the start. Added automated test.

[Jared Jurkiewicz]

(In [21293]) Fixing selection restore issue caused by 10678, broke selection restore on IE. refs #10678

[Jared Jurkiewicz]

(In [21294]) Didn't mean to leave prettyprint enabled. refs #10678

[Jared Jurkiewicz]

(In [21295]) Fixing selection restore issue caused by 10678 futher, need to cleat on mousedown too. refs #10678

[Jared Jurkiewicz]

(In [21316]) Fixing regression triggered by the moveCursorToStart added by Doug. The error occurs in the case of the first mouse click to set position being ignored if a plugin that requires position is then immediately used. Need to clear that flag on position clicks. \!strict refs #10678