On Mon, Aug 30, 2010 at 09:57:11AM +0200, Zsbán Ambrus wrote: > Dear Marc, > > You might remember our discussion about that a backspace output when > the cursor is after the last column, urxvt moves the cursor to the > last but one column. You claimed not only that this conforms to > standards but also that most vt100 terminal emulators work this way: > > On Mon, Sep 28, 2009 at 11:58 AM, Marc Lehmann <[email protected]> wrote: > > Well, it fails with wide characters, for example, and basically all vt100 > > emulators, which is the most common terminal type. > > I only have one new observation to this. Gnu screen which is a > vt100-like terminal, also has backspaces wrapping from the beginning > of a line to the previous line, and it treats backspace from after the > last column the other way: it just moves the cursor to the last > column. I tested this with at least Screen version 4.00.03jw4 (FAU) > 2-May-06 (from debian lenny linux amd64).
Not only GNU screen - I noticed that xterm also does this, but only when "Enable Reverse Wraparound" is selected in its settings (xterm -rw, or set *VT100.reverseWrap: true). With both auto wraparound (-aw, autoWrap) and reverse wraparound enabled the backspace received when the cursor is "after the last column" (displayed as in the last column) just moves the cursor to the real last column, not to the position before it. And with such backspace handling the BS-space-BS erasing used by the terminal layer works perfectly even for multiline input (at least when no wide characters are entered) - the cursor position does not get out of sync with the input buffer contents, and there is no garbage left on screen.
signature.asc
Description: Digital signature
_______________________________________________ rxvt-unicode mailing list [email protected] http://lists.schmorp.de/cgi-bin/mailman/listinfo/rxvt-unicode
