On Fri, 12 Jul 2002, Rick Moen wrote: > Quoting Thomas Dickey ([EMAIL PROTECTED]): > > > the same as before: the 'screen' terminfo which has no 'bce' capability. > > So, a couple of things about that, if you don't mind: > > 1. Perhaps you wouldn't mind explaining the relevance. _Is_ Background > Color Erase support in one's terminfo setup essential for screen to > correctly support X11 copy/paste (without getting spurious right-padded > spaces? Is this a known requirement and a known failure mode in its > absence?
It's a matter of side-effects. bce says that when doing an erase, the terminal doesn't change the background color. (n)curses and other applications use this information to decide when they don't have to repaint the background using spaces, i.e., filling using the current background color setting. xterm uses the explicit writes in contrast to erases to decide what parts of the screen contain text which can be selected. > Please note that I've tried to research this in the docs for both screen > and mutt, and not found anything. Is this documented somewhere, and I > just missed it? > > 2. If so, why did all this use to work on default Debian (circa 2.2 > release) setups, and then suddenly cease to work upon upgrading _mutt_, > at some point? Please note that I changed no terminal settings, during > that period. I've been using mutt under screen for many years, and all > I did was incrementally upgrade both precompiled Debian packages using > apt-get. probably the version of screen changed. At one point it didn't pay much attention to the bce feature, iirc. I still have a note in one of my to-do lists from the beginning of 2000 that screen had something hardcoded to assume that $TERM set to xterm-color implied that it supported bce. That may be the source of your confusion. It was incorrect, and presumably has been fixed. (I've sent fixes a several times to screen's maintainers, who have ignored them, btw - my recollection is that the only time I get an email response is when I'm not supplying a patch ;-) > 3. Three days ago, when I attempted to enable BCE on someone else's > advice, you may recall that I created ~/.terminfo/s/screen as a symlink > pointing to /usr/share/terminfo/s/screen-bce . On Debian, the latter is > itself a symlink to /etc/terminfo/s/screen-bce, a compiled terminfo > entry. Running strings on that binary yields the following header: > > screen-bce|VT 100/ANSI X3.64 virtual terminal with bce > > So, are you saying this terminfo entry purports to include BCE support > but is mistaken? Or that terminfo is otherwise defective? Or that > terminfo BCE support is there but screen is failing to use it, or what? It doesn't mention bce. bce is a boolean flag, and infocmp would print that on the first few lines. The names are ordered according to type. -- T.E.Dickey <[EMAIL PROTECTED]> http://invisible-island.net ftp://invisible-island.net
