On Fri, Dec 22, 2006 at 03:04:53PM +0100, Vincent Lefevre wrote: > On 2006-12-22 06:27:26 -0500, Thomas Dickey wrote: > > If there's no new information here, there's nothing to fix (the > > fixes are in ncurses 5.5, and there's no way other than by breaking > > the related fixes for luit to make ncurses 5.4 work as you want). > > The problem is that Debian/stable still provides ncurses 5.4. > So, there must be a fix for Debian. > > Also, concerning Mac OS X, though I have ncurses 5.5 installed (via > MacPorts), the problem is that several programs are written for curses > (instead of ncurses), and ncurses 5.5 doesn't provide compatibility > links. I've already reported a few bugs for MacPorts and written a few > fixes, e.g. > > http://trac.macosforge.org/projects/macports/ticket/11157 I've seen some confusing comments about Mac OS X "curses" versus "ncurses", but am left with the impression that it's still ncurses (in a different directory, etc, but still the same code). Even still, 5.4 is nearly three years old, and there's a year-old 5.5 release which Mac OS X should be using. > > Also note (as you already did) that there's a known workaround (use the > > terminal description that uses the single-character ^N and ^O rather than > > \E(B, etc. > > But this time, these characters break "less"! > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=399848 That's odd, since "less" has always been a termcap application. Perhaps it's a regression, e.g., changes by someone using a terminfo underneath. > And I don't know if this is related to changes in ncurses: > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=403612 Since "less" uses only the termcap interface, it would only be related to changes in that one area. Essentially the changes to sgr0 were made necessary by a long-ago equating of terminfo "sgr0" and termcap "me". They're not identical: a) several years ago, I made changes to allow termcap applications to assume that "me" (sgr0) did not reset line-drawing since termcap applications generally did not support that feature. b) that change only worked for single-bytes - and the change I noted yesterday was a fix for that. The fix was needed to allow using multi-byte \E(B, etc., to support luit. c) further changes past that point were refinement (improving the interaction with applications that had not been fixed with (b)). -- Thomas E. Dickey http://invisible-island.net ftp://invisible-island.net
Attachment:
pgpFKIWUJvHjn.pgp
Description: PGP signature