Hello,
I have encountered a situation where I observe terminal corruption when
using emacs (either emacs-nox or emacs -nw). The corruption can be
demonstrated easily using isearch-forward for a single character which
appears multiple times in the current buffer, and repeatedly searching
to cause a wrap back around to the start.
Most commonly, the corruption involves the majority of the terminal
undergo a linefeed, but with dynamically highlighted sections being
rendered in the correct absolute location on the terminal. Less often,
the corruption is limited to the text in the minibuffer alone. A
complete buffer redraw (C-l) shows that the buffer contents is correct,
and this temporally resolves the corruption.
I have several baremetal Debian systems (6, 7 and 8) and 3 Debian
systems (7, 8, 8 with no desktop) as virtualbox guests on my laptop
(windows sadly). All 64bit.
Each of the virtuabox guests have absolutely no terminal corruption when
emacs is run from a real terminal (tty1,2,etc) but exhibit the
corruption when run from a pseudo terminal. The baremetal systems show
no corruption at all, even in pseudo terminals.
Corruption on the ptys occur in all cases I can construct; console
emulators (including gnome-termainal, mate-terminal, xfce-terminal, and
plain xterm), sshing in from outside, or being on a tty and running
emacs inside screen.
I have experimented in various ways by trying to make the environments
identical (as reported by `env`), making the tty setttings identical (as
reported by `stty -a`), and even by copying the 'good' (baremetal) xterm
terminfo and using strace to verify that the new terminfo is being
loaded. In addition, I have tried with and without virtualbox guest
utilities.
I was able to control one difference. using TERM=linux in a pty (to
match TERM= in a tty) removes the linefeed corruption, but still leaves
corruption in the minibuffer when the amount of text would reduce. Is
it possible that all the corruption is just an issue with blanking the
terminal, or subportions thereof?
This does not obvious appear to be a bug with emacs itself (23 or 24),
or anything about a particular version of Debian (both 7 and 8 affected
equally, with 3.2 and 3.13 kernels respectively). The only common
factor is pseudo terminals inside virtualbox guests, which is quite strange.
I am now out of ideas to continue. Does anyone have any suggestions for
what this issue might be, or how I can debug the issue further?
Thanks in advance,
~Andrew
--
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Archive: https://lists.debian.org/549f5bc4.1090...@cam.ac.uk
