aborted zsh completion not fully cleared

[octoploid]

log.tar.zip
Start tmux -> dmesg -> type man m[tab]

type [backspace] to abort:

[nicm]

Are you sure you have 7a4c66b?

Does this show the same problem (compare inside and outside tmux):

clear;dmesg;printf '\033[3;1H\033[J\033[10;1H'

[octoploid]

I have 7a4c66b and no, your command line doesn't show any problems.

[nicm]

Does this also happen in xterm? Make sure TERM is the same outside tmux in both terminals.

[nicm]

Never mind, this is the same bug in Konsole. I thought you couldn't scroll more than the scroll region, but it looks like you can't scroll the whole scroll region either. Compare this in xterm:

clear;seq 100;printf '\033[3;20r\033[18S\033[r';sleep 10

And:

clear;seq 100;printf '\033[3;20r\033[17S\033[r';sleep 10

With a scroll region of 3-20 we need to scroll 18 lines to remove the whole thing. 17 is not enough.

So 7a4c66b is wrong, and I am going to revert it.

tmux needs a working indn if it is specified in terminfo. What did the Konsole guys say to your bug report?

You should be able to work around this for now by doing:

set -as terminal-overrides ',*:indn@'

[octoploid]

Yes, it also happens in xterm.

[octoploid]

Commit cb5fcb3 doesn't fix this issue for xterm.

[nicm]

Show me logs from using xterm with latest master please (only need the server log).

[octoploid]

Sorry, false alarm (had an unnoticed tmux instance running in the background).
Xterm is fine now.

[nicm]

OK cool. I don't think there is a practical way to detect this bug, I think it will need to be fixed in Konsole.

[octoploid]

I would be good if you could open a Konsole bug yourself, because I don't understand the issue sufficiently:
https://bugs.kde.org/enter_bug.cgi?product=konsole

[octoploid]

Ah, found the konsole bug: https://bugs.kde.org/show_bug.cgi?id=379318

[nicm]

Yes I think this bug report is fine. If you can figure out how to build Konsole you could try this, which copies the check in Screen::scrollDown() to Screen::scrollUp():

diff --git a/src/Screen.cpp b/src/Screen.cpp
index d456d67..02c1e4c 100644
--- a/src/Screen.cpp
+++ b/src/Screen.cpp
@@ -868,7 +868,9 @@ QRect Screen::lastScrolledRegion() const

 void Screen::scrollUp(int from, int n)
 {
-  if (n <= 0 || from + n > bmargin) return;
+  if (n <= 0) return;
+  if (from > bmargin) return;
+  if (from + n > bmargin) n = bmargin - from;

   _scrolledLines -= n;
   _lastScrolledRegion = QRect(0,tmargin,columns-1,(bmargin-tmargin));

[nicm]

Actually that was a very old tree, try this:

diff --git a/src/Screen.cpp b/src/Screen.cpp
index 90671829..fd8e5c3c 100644
--- a/src/Screen.cpp
+++ b/src/Screen.cpp
@@ -752,7 +752,12 @@ QRect Screen::lastScrolledRegion() const

 void Screen::scrollUp(int from, int n)
 {
-    if (n <= 0 || from + n > _bottomMargin) return;
+    if (n <= 0)
+        return;
+    if (from > _bottomMargin)
+        return;
+    if (from + n > _bottomMargin)
+        n = _bottomMargin - from;

     _scrolledLines -= n;
     _lastScrolledRegion = QRect(0, _topMargin, _columns - 1, (_bottomMargin - _topMargin));

[octoploid]

Thanks. I have created: https://git.reviewboard.kde.org/r/130133/diff/1#index_header

(there was an off by one error in your patch, which caused exactly the same issue that I've described above)

diff --git a/src/Screen.cpp b/src/Screen.cpp
index 90671829bf9b..45b8da0ab0e9 100644
--- a/src/Screen.cpp
+++ b/src/Screen.cpp
@@ -752,7 +752,12 @@ QRect Screen::lastScrolledRegion() const
 
 void Screen::scrollUp(int from, int n)
 {
-    if (n <= 0 || from + n > _bottomMargin) return;
+    if (n <= 0)
+        return;
+    if (from > _bottomMargin)
+        return;
+    if (from + n > _bottomMargin)
+        n = _bottomMargin + 1 - from;
 
     _scrolledLines -= n;
     _lastScrolledRegion = QRect(0, _topMargin, _columns - 1, (_bottomMargin - _topMargin));

[octoploid]

Not a tmux bug. Lets close this issue.

[lock]

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.