Issue #2671

Issue #2665

It doesn't need it

Now nested operations from different editors end at the same time

Issue #2678

Looks at token types to determine whether the quote would actually open a string

Issue #2657

Issue #2663

To make it less likely that an error will corrupt its value and
break all further signalLater events.

Issue #2626

Issue #2659

Also updated select block to return selectionStart.
Operators from now on will stop using curStart and curEnd, and rely on the current selection instead. The selection will be generated in evalInput if needed.

Currently, if you perform the uncomment command with the cursor outside a single-line block comment, it will uncomment that block comment. This is unfortunate behavior for cases where the block comment is used to describe the parameter to a function like "foo(/* index */ 0)". It's also inconsistent: if you have a selection and a single-line block comment after the end of the selection, uncomment() will do nothing, however if you have a selection with a single-line block comment *before* the selection, it will line-comment out all the lines of the selection instead.

This also isn't a complete fix, but significantly reduces the scope of the problem. For example if you have "/* foo */ bar /* baz */", uncomment() still does the wrong thing when the cursor is on foo or baz.

http://lists.w3.org/Archives/Public/www-style/2014Jun/0312.html

Lost negation operator when converting to new IE flags in bbefd22a.

Issue #2635

Issue #2644