determine when isWrapped should be reset to false

We set isWrapped to true for a buffer line, once the text overflows right side in wrapAround mode in InputHandler.print.

We should define circumstances, when this settings has to be reset to the default value of false. Currently there are a few bug reports associated with this: #1710, #1713 and a partial fix in #1720.

Proposed behavior:

• keep isWrapped as long as possible
• reset it to false once the following happens:
  • a horizontal resize event: Currently we dont support reflow on resize, therefore isWrapped will not reflect the correct line wraps anymore upon horizontal resize. It should simply set all isWrapped to false. Would cover #1713.
  • some cells of a line got changed by cursor repositioning: This is a tricky one since the intention of changing the old content is not clear. I think as soon as anything gets changed by repositioning the cursor the terminal is used in a more advanced manner and we should also reset isWrapped to false making them single content lines. Would cover #1710 and #1720.

Up for discussion @Tyriar, @mofux, @bgw

[Tyriar]

a horizontal resize event: Currently we dont support reflow on resize, therefore isWrapped will not reflect the correct line wraps anymore upon horizontal resize. It should simply set all isWrapped to false. Would cover #1713.

I don’t think we should touch isWrapped here, the fix I had in mind of #1713 was to special case it until we have reflow in place.

[egmontkob]

I think as soon as anything gets changed by repositioning the cursor the terminal is used in a more advanced manner and we should also reset isWrapped to false

I’m afraid it might be too much.

The use case that occurs to me is bash (readline), zsh and friends. The nice user-friendly behavior is if the command line being edited isWrapped (all lines except the last). There are various ways of editing a part of a line that doesn’t require a full repaint, e.g. Ctrl+T for transposing two adjacent characters. I checked bash-4.4 now and it doesn’t reprint the entire line, only the two letters that swap places. Yet the command being typed should remain isWrapped.

In VTE we made it behave roughly like xterm does (roughly == I’m not aware of any difference but we didn’t do a thorough check so there easily might be), and the cases where the flag is reset are surprisingly few. Clear screen (or a part of the screen, then obviously for those lines only), clear to end of line, delete char (but surprisingly not at insert char), new line appearing due to scrolling (perhaps with a scroll region defined). Taking a quick look at our source code, I think this is all.

xterm doesn’t do rewrapping either, but doesn’t alter this flag on horizontal resize. A double/triple click still selects multiple lines at once. The newly added cells on the right are “unused” ones, skipped when copy-pasting.