This would be one application for a generic region layout. Panes and status
lines and parts thereof could be discernable areas with custom styles and mouse
click actions. Sort of what curses already allows, IIRC.
On May 25, 2015 2:27:26 AM CEST, Daniel Cotton
Applied, but without the shift part.
Thx!
The shift part however is necessary in those cases mentioned where the
program has broken mouse support..
Because it gives all CLI programs (even bash history) intuitive mouse
support without updating/fixing every single one of them. There are
and a small pane.
#Regards.
From ce524403411ca19f50cb6af65c0994cd7e762776 Mon Sep 17 00:00:00 2001
From: Marcel Partap mpar...@gmx.net
Date: Sun, 13 Apr 2014 00:14:07 +0200
Subject: [PATCH 2/7] Re: m-scroll in choose-*: set to 1 by default?
---
window-choose.c | 3 ++-
1 file changed, 2 insertions
Went with this, I think line-at-a-time is fine for choose mode, most of
our lists are pretty short.
Objection: please take the patch I sent for this (13/04/2014 00:56) as
this one looses flexibility of using CTRL to speed up.
The assumption that choose-lists are small may be true for YOUR usage
.
#Regards.
From 419a8102b6308477f62c54278eab77d034c8e8d3 Mon Sep 17 00:00:00 2001
From: Marcel Partap mpar...@gmx.net
Date: Sun, 13 Apr 2014 00:31:10 +0200
Subject: [PATCH 1/2] Don't wrap mouse scrolling in choice-mode lists v2
Feels somewhat more natural. Objections?
---
window-choose.c | 9
---
window-choose.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/window-choose.c b/window-choose.c
index e75858e..9406b9c 100644
--- a/window-choose.c
+++ b/window-choose.c
@@ -724,7 +724,8 @@ window_choose_mouse(
u_inti, idx;
Feels somewhat more natural. Objections?
---
window-choose.c | 9 ++---
1 file changed, 6 insertions(+), 3 deletions(-)
diff --git a/window-choose.c b/window-choose.c
index 9406b9c..8f1ab5c 100644
--- a/window-choose.c
+++ b/window-choose.c
@@ -726,10 +726,13 @@ window_choose_mouse(
Rationale:
· remove _MASK_ from SHIFT/META/CTRL for consistency with Key modifier bits
KEYC_ESCAPE/KEYC_CTRL/KEYC_SHIFT
· rename MOUSE_MASK_BUTTONS to MOUSE_MASK_BUTTON: only receiving single button
events
· comment rename button masks = event masks: drag and wheel are more events
than they are
No issues found in weeks of heavy use, please test anyway.
---
input-keys.c | 14 ++
1 file changed, 14 insertions(+)
diff --git a/input-keys.c b/input-keys.c
index f092f97..1df8e98 100644
--- a/input-keys.c
+++ b/input-keys.c
@@ -204,6 +204,20 @@ input_mouse(struct window_pane *wp,
This breaks consistency between different modes; does SHIFT+scrolling
not suit you well enough? That does single-line-scrolling but is not
documented yet, so you may have missed it.
#Regards/MPartap
--
Put Bad Developers
(+ some other changes regarding mouse implementation)
- is u_int for i ok?
- was there a valid reason to use bit shifts for MOUSE_EVENT_*
constants?
- ...
---
input-keys.c| 18 +-
server-fn.c | 1 +
tmux.h | 23 +--
tty-keys.c | 27
Actually, it doesn't need a server_ function.
Then again, it doesn't compile for some obscure reason.
Huh?
From 92a2e3db2cd2e8fabb01762bec678c3868a46e11 Mon Sep 17 00:00:00 2001
From: Nicholas Marriott nicholas.marri...@gmail.com
Date: Wed, 26 Feb 2014 23:04:23 +
Subject: [PATCH] Implement
Hmm was working for me.
Nah I meant the revised version I sent in reply.. here's the latest. It
fails to compile however (help?).
From 2e3633271798d9052953d23e5317812bdc32c015 Mon Sep 17 00:00:00 2001
From: Nicholas Marriott nicholas.marri...@gmail.com
Date: Wed, 26 Feb 2014 23:04:23 +
It fails to compile
more exactly:
depbase=`echo tty-term.o | sed 's|[^/]*$|.deps/|;s|\.o$||'`;\
gcc -DPACKAGE_NAME=\tmux\ -DPACKAGE_TARNAME=\tmux\
-DPACKAGE_VERSION=\1.10\ -DPACKAGE_STRING=\tmux\ 1.10\
-DPACKAGE_BUGREPORT=\\ -DPACKAGE_URL=\\ -DPACKAGE=\tmux\
-DVERSION=\1.10\ -DSTDC_HEADERS=1
I usually find it helpful to compile with clang in these situations
(CC=/usr/bin/clang and CXX=/usr/bin/clang++). The error messages are
usually much more helpful.
Sheesh that was helpful indeed THANKS :)
# CC=clang ./configure --enable-debug
...
depbase=`echo tty-term.o | sed
Thx, found already out by using clang (make --debug was not useful)..
It does when you realise term.h has #define lines
which is quite a naughty name choice imho.
#Regards.
--
Flow-based real-time traffic analytics
---
input-keys.c| 17 +
options-table.c | 7 +++
tmux.1 | 23 +++
tmux.h | 5 +
window-choose.c | 19 ++-
window-copy.c | 12 +---
6 files changed, 79 insertions(+), 4 deletions(-)
diff --git
[ Now why did git send-email munge my memo? .. anyway...]
Here's the 3rd iteration, including a wee man page section.
The code duplication sucks.
Comments welcome.
#Regards.
--
Flow-based real-time traffic analytics
Can the duplicated code not be refactored into a shared function? It
looks like you could pass in m and return an integer representing the
number of key events to send?
Definitely, but I'm not quite sure where to put it...
Gaaah! I just freaked myself out [..] guess I must have done something wrong
Happens :D
Final quirk: I'm not seeing any altered behaviour when I hold down the
shift, meta, or alt keys: from what I see in the code, aren't they
supposed to alter the scroll rate?
In which state? copy-mode,
: Marcel Partap mpar...@gmx.net
Date: Mon, 17 Feb 2014 22:02:14 +0100
Subject: [PATCH] Implement simple mouse wheel emulation.
---
input-keys.c | 15 +++
tmux.h | 5 +
2 files changed, 20 insertions(+)
diff --git a/input-keys.c b/input-keys.c
index 7582a63..5c5e105 100644
Thanks for your quick review good catch, please try again with
attached revision. c;
#Regards!
P.S.: [some patch NIH broke the command history in alternate screen..]
From 48c5d8b8bf332b3f67f77154561a36027e013198 Mon Sep 17 00:00:00 2001
From: Marcel Partap mpar...@gmx.net
Date: Mon, 17 Feb 2014
* With mouse-mode=on, vim scrolls as expected with the mouse wheel, but
less does not: it enters copy mode.
Can only reproduce with less --no-init (no alternate screen).. does
# env|grep LESS
show any special flag settings?
* With mouse-mode=off, vim and less both scroll as expected.
What about
Hi,
On 1.7 when I had 2 panes the whole pane line was colored but now with
1.8 only the half, this difference made be to believe that 1.8
introduced a bug. But after your mail I realized that this is actual a
very useful feature.
Interesting... I was a bit confused by this as well, until I
---
screen-redraw.c | 77 -
1 file changed, 27 insertions(+), 50 deletions(-)
diff --git a/screen-redraw.c b/screen-redraw.c
index 6131907..1af1bb1 100644
--- a/screen-redraw.c
+++ b/screen-redraw.c
@@ -26,8 +26,6 @@ int
Rename
screen_redraw_cell_border1() = screen_redraw_get_border_rel()
screen_redraw_cell_border() = screen_redraw_cell_is_border()
screen_redraw_check_cell() = screen_redraw_get_cell_type()
screen_redraw_check_active() = screen_redraw_check_active_pane_indicator()
---
screen-redraw.c | 67
---
examples/tmux.vim | 2 +-
options-table.c | 9 +
screen-redraw.c | 27 ++-
tmux.1| 5 +
tty-term.c| 2 +-
5 files changed, 38 insertions(+), 7 deletions(-)
diff --git a/examples/tmux.vim b/examples/tmux.vim
index
---
screen-redraw.c | 2 ++
tty-acs.c | 64 -
2 files changed, 34 insertions(+), 32 deletions(-)
diff --git a/screen-redraw.c b/screen-redraw.c
index 1af1bb1..9e387c3 100644
--- a/screen-redraw.c
+++ b/screen-redraw.c
@@ -44,6 +44,7
about.. UI overlays / windows?
Date: Sun, 14 Jul 2013 15:33:50 +0200
From: Marcel Partap mpar...@gmx.net
To: tmux-users@lists.sourceforge.net
On 14/07/13 15:33, Marcel Partap wrote:
Salut,
recently it occurred to me several times (f.e. when looking up list-keys
or choose-tree, also while using
An overlay layer above the current window/pane/border/status line layer.
To display information overlays, option/selection dialogs, popup
menus (f.e. mouse right click on a window = split, kill, make
sticky, rename...)
One very use case is a character selection dialog for send keys,
allowing
I can explain how everything fits together if anyone wants to have a go.
Lol I so would like to because I have a lot of use cases right there in
my mind but am simply not adept enough of a coder to get it done in a
justifiable amount of time c;
I've had this problem before. Sometimes 'reset' doesn't fix it because
the binary data includes the magic tmux escape code to set the window
name. I have to go in and explicitly change the window name back, and
then it works fine.
Splitting the window, breaking off the f0obared pane and
The statusline indeed changes, when i start xterm it
shows - for a moment then hides for several seconds...
then starts this again forever.
..very likely that this is because it takes powerline longer to generate its
output than what your status interval is set to. You can test that by
We all have root permission. How can I dissable
others attach to my session?
Use a socket-name other than default. F.e. :
tmux -L $(USER)
That'd prevent accidental session take-over.
#regards
--
See everything from the
tmux -L $(USER)
Uhhm 'tmux -L ${USER}' that is. Or simply '-L $USER'. Or '-L $RANDOM'...
--
Diese Nachricht wurde von meinem Android-Mobiltelefon mit K-9 Mail gesendet.
--
See everything from the browser to the
[...]
Thomas, you're [...]
Ok you crossed a thick red line right there. Would you bloody please
stop insulting ANYONE on ANY public opinion forum EVER AGAIN?
Please take some time far off a keyboard and reflect your demeanour. How
you dare spewing rants of disrespect at people whose hard
Dear Jason,
you are still not far enough away from the keyboard.
On 27/04/13 00:08, Jason Timrod wrote:
I am voicing [...] shit, [...] and it's not good enough.
--
Try New Relic Now We'll Send You this Cool Shirt
New
On Tuesday evening -- I'm planning to release tmux 1.8.
Any questions, please let me know.
Yes, would you be so kind and delay the release a couple of hours. Only
just took notice of it and imho some of the patches I sent in a year ago
and then again couple of months back should really really
Oh and the recent merge does look ugly indeed - cherry-picking is a much
better way to deal with situations like that
...further diving into the git history, I dismiss my previous
statement.. cherry-picking from obsd-master is obviously not a real
option.. but is there no way to merge without
pager. Because mouse scrolling is for wimps only
anyways. Huh?
#Regards.
From 0a05c60331c99b87cef31ea4117f8a0144cb0422 Mon Sep 17 00:00:00 2001
From: Marcel Partap mpar...@gmx.net
Date: Sun, 24 Jun 2012 23:17:06 +0200
Subject: [PATCH 3/7] Rename mode-mouse option to mouse-copy-mode
and remove
..ok that was meant to be a reply to
20130326160843.ga22...@yelena.nicm.ath.cx sorry^^
--
Own the Future-Intelreg; Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest.
Compete for
...
isn't going to happen now, since I'm about
to cut a release for 1.8.
Uhm ok.
But adding in your changes now -- even by delaying 1.8 by a few days --
simply is not an option.
Because?
We would need much more time to ensure new code
isn't going to introduce bugs
Much more time? The code
You might want to disable ALL mouse support and that has to be
possible. In fact, it has to be the default.
Ok if so, then it needs at least one more option in addition to
mouse-copy-mode.. mouse-choose or something.
We could split it up into a few options but what?
What? Clarity and
I don't have a scroll wheel here but
I'm happy without an option and just to hardcode 3 lines
Nah, that's so not good enough.
And having used various incarnations of my patch for a considerable
amount of time, the flexibility of a variable setting and the modifier
keys really is a true asset for
First usage instance is the renaming of the mode-mouse option.
Hmm. I am not sure I'm too keen on this approach.
a) Please elaborate.
b) Do you have a better idea to warrant config file backwards compability?
c) More opinions needed!
What happens when/if
mode-copy-mouse also needs renaming
+#define CELL_BORDERS xqlkmjwvtun~ /* │─┌┐└┘┬┴├┤┼· */
+#define CELL_MARKERS +, . - /* →← ↑ ↓ */
And if I don't have a font capable of displaying the characters in the
comments?
Then you should upgrade your text editing system of choice to be able to
handle 21st century
by picking up the process group leader from the terminal fd
via tcgetpgrp() instead of using the direct pane child process.
Why? I mean, what problem is this solving? You don't mention it in your
commit message.
Nested subshells and su invocations. Consider this:
# root key bindings
and remove condition to enable mouse in choose mode.
This time, no backwards compatible mercy. RTFM!
---
input-keys.c|2 +-
options-table.c |8
server-client.c |2 +-
tmux.1 | 11 +--
window-choose.c |3 +--
window-copy.c |2 +-
window.c
by picking up the process group leader from the terminal fd
via tcgetpgrp() instead of using the direct pane child process.
This correctly picks up the working directory for nested subshells.
(works on linux, other platforms untested)
---
cmd.c |2 +-
format.c |2 +-
Scrolling mouse wheel while shift modifier key is held generates
fake cursor key events. With ctrl modifier added, speed is multi-
plied by the value of mouse-scroll-speedup variable (default 3).
---
examples/tmux.vim |2 +-
input-keys.c | 18 --
options-table.c |
---
screen-redraw.c |4 ++--
tty-acs.c | 64 +++
2 files changed, 34 insertions(+), 34 deletions(-)
diff --git a/screen-redraw.c b/screen-redraw.c
index 75f7238..7ffad30 100644
--- a/screen-redraw.c
+++ b/screen-redraw.c
@@ -41,8
Scrolling mouse wheel while shift modifier key is held generates
fake cursor key events. With ctrl modifier added, speed is multi-
plied by the value of mouse-scroll-speedup variable (default 3).
---
trunk/examples/tmux.vim |2 +-
trunk/input-keys.c | 15 +--
First usage instance is the renaming of the mode-mouse option.
---
trunk/cmd-set-option.c |3 ++-
trunk/options-table.c | 55
trunk/tmux.h |1 +
3 files changed, 58 insertions(+), 1 deletion(-)
diff --git
---
trunk/examples/tmux.vim |2 +-
trunk/options-table.c |5 +
trunk/screen-redraw.c | 38 +++---
trunk/tmux.1|4
trunk/tty-term.c|2 +-
5 files changed, 38 insertions(+), 13 deletions(-)
diff --git
by picking up the process group leader from the terminal fd
via tcgetpgrp() instead of using the direct pane child process.
(works on linux, other platforms untested)
---
trunk/cmd.c |2 +-
trunk/format.c |2 +-
trunk/osdep-aix.c |2 +-
trunk/osdep-darwin.c
---
trunk/screen-redraw.c |4 ++--
trunk/tty-acs.c | 64 -
2 files changed, 34 insertions(+), 34 deletions(-)
diff --git a/trunk/screen-redraw.c b/trunk/screen-redraw.c
index 75f7238..4d4899d 100644
--- a/trunk/screen-redraw.c
+++
---
trunk/input-keys.c|2 +-
trunk/options-table.c |8
trunk/server-client.c |2 +-
trunk/tmux.1 | 11 +--
trunk/window-choose.c |3 +--
trunk/window-copy.c |2 +-
trunk/window.c|2 +-
7 files changed, 14 insertions(+), 16
..
Ok then how about supporting both in code, while changing the man page
to 'mouse-copy-mode' ?
#regards...
--
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat
So Tmux always forwards mouse input then?
Meaning, it doesn't interfere with it, yes.
#regards c:
--
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape
G, what the rush^^
have been working on this and here's what i have in tree right now.
These are somewhat untested, will test and report back later... C:
From a824632d2fb794fbc0e64d47fc15ef6e9e219b47 Mon Sep 17 00:00:00 2001
From: Marcel Partap mpar...@gmx.net
Date: Wed, 20 Jun 2012 15:48:27
You need to check the logic for lines in input_mouse() -- holding down
ctrl and scrolling down all the way to the bottom of the window causes the
server to crash. :)
ANY scrolling caused crashing. Because fatalx(missing option);
Well I put the option inbetween the other mouse-* options - in
server to crash. :)
uncrashed.
From 7f588c64242b3c1d8a31d3bc6aaa1b590027b776 Mon Sep 17 00:00:00 2001
From: Marcel Partap mpar...@gmx.net
Date: Thu, 21 Jun 2012 12:01:55 +0200
Subject: [PATCH 4/4] Cleaned up implementation of mouse wheel scrolling
emulation.
---
trunk/examples/tmux.vim
Be as it may, what went in is rather only fixing odd behaviour..
What exactly is the functionality you are missing Fabio?
--
#regards/marcel
--
Live Security Virtual Conference
Exclusive live event will cover all the
Hi there fellow h4x0rz ;)
just wanted to have a comment on these, will finalize them after my exam
end of the month...
tmux-quiet-set-option.patch:
still useful, my use case is setting option window-status-fg when
splitting a window with su -l [different user]
Hi Nic,
Is this finished or are you still working on it?
Finished. The last patch i posted is what i use at the moment.
#regardz | marcel C;
--
Keep Your Developer Skills Current with LearnDevNow!
The most comprehensive
Does anyone have this working and can offer advice? Or anyone can
think of something else to try?
Please try the patch i posted yesterday.
#regards.
--
Ridiculously easy VDI. With Citrix VDI-in-a-Box, you don't need a
On 24/12/11 09:27, Nicholas Marriott wrote:
Yes, not intended. Fix applied, thanks!
Uhhm.. well. Rerolled my previous patch (and converted the time_diff
function into a #MACRO). Another one here, don't exit mouse-mode for
mode-mouse=copy-mode when reaching bottom.
#regards/marcel.
---
Fellow tmuxers,
at LAAAST! It is completed! ^^
There might be a few touch ups here and there that could be done but all
in all, my mouse overhauling effort has concluded with attached patch.
Changes:
- For panes not in xterm mouse tracking mode, 'emulate' wheel scrolling
by sending cursor
On 06/12/11 09:06, Nicholas Marriott wrote:
I guess ncurses just do this with a queue and consuming them into to
build a mask which you could probably do too, but I'd have to go poke
around in the ncurses code to see for sure.
that's a good idea btw. Here.. will look later. #best
Thanks for your replies, I'll wait to see if a version of this patch will be
included in tmux, otherwise, I'll patch myself.
The patch as posted works fine as is, have not had any problems. It has
some debug code still in that i will rip out ASAP, but please test if it
works for you.
..just for reference, from konsole/src/ProcessInfo.cpp, includes omitted:
virtual bool LinuxProcessInfo::readCurrentDir(int pid)
{
QFileInfo info( QString(/proc/%1/cwd).arg(pid) );
const bool readable = info.isReadable();
if ( readable info.isSymLink() )
phew, mind twisting TAILQ_* macros..but it works now. The problem
you described is gone with this, and also this patch allows for complex
mouse click patterns (like kill window on triple middle button click
then shift left click within one second).. that needs a config syntax
though. and
Don't shell out to date here -- the status lines get run via
strftime() -- so just use the format specifiers for that directly and
your load-average will go away, hopefully.
Yes, BUT i've had the problem disappear for other programs forked from
the status bar aswell. There's something wrong
This is weird
indeed ^^
Can you send tmux info output?
Yes we can C:
tmux 1.6, pid 8196, started Sat Sep 17 04:32:41 2011
socket path /tmp/tmux-0/default, debug level 0
system is Linux 3.1.0-rc4-00131-g9e79e3e #27 SMP PREEMPT Thu Sep 1 12:05:51
CEST 2011 x86_64
configuration file is
Hi there :)
This is a strange thing, sometimes tmux totally pwns one CPU, just
noticed it has been consuming bloody 30hours of precious 2.8GHz
processor time that should have been going to BOINC *grrr* ^^
anyways, a crude statistical analysis after detaching all clients
sleep 10s killall -v
I don't wanna show tmux split lines,how?
set-option pane-border-fg black
set-option pane-active-border-fg black
--
BlackBerryreg; DevCon Americas, Oct. 18-20, San Francisco, CA
http://p.sf.net/sfu/rim-devcon-copy2
this is a great idea. i've been interested in making tmux tablet friendly
Heh well totally (:
in various ways, and my current solution (dwm + xdotool + shell scripts that
call tmux) is totally hack. i'd like to see this even expanded a bit so that
one could in their .tmux.conf bind functions
...With this, a double click on the status line (any part not belonging
to the window list) creates a new window.
Tricky stuff. Now i know much more about tmux debugging ^^
As i couldn't find a way to receive mouse button AND modifier, window
closing on CTRL+double click is yet missing.
haven't got time to look into this really, but it is on the pile in my
inbox so if nobody else does I'll get to it after this week
Ok i have wasted couple of hours investigating this and conclusion is as
follows.
If any mouse-interactivity feature (f.e. mouse-select-pane) is turned on
(and, in
/src/bulk/console/tmux/repo-cvs/github.com-ThomasAdam-tmux/tty.c:1006:
undefined reference to `__b64_ntop'
just hit the same, apparently there's a -lresolv missing in makefile LIBS...
--
vRanger cuts backup time
I don't follow, what is the problem? If you turn on mouse-select-pane it
eats the scroll wheel button events?
no, it seems to alter them. ncmpcpp reacts on wheel events, less not.
I think it would be better for scroll wheel to select panes not
windows.
The patch only enables mouse wheel
Hi Nicholas et al.,
seeing that the previous issue (excessive redraws) seems to be solved (have not tested but patch seems clear), i want to bring up
another minor issue. Mouse scroll wheel events are only correctly passed if mouse-select-* is off.. while ncmpcpp can handle mouse
wheel input
On 05/10/10 16:45, Nicholas Marriott wrote:
true still forks a child.
What are you setting status-left and right to again?
set -g status-left [ #[fg=green]#H #[fg=white]]
set -g status-right [ #[fg=magenta]#(true)#[default] ][ #[bold]#(true)
#[fg=blue]#(true)#[default] ]
On 07/10/10 22:05, Nicholas Marriott wrote:
So with this the problem is that although you do not see a growing
number of true programs running with ps -aelf, you do see tmux using
an increasing amount of memory and CPU in eg top?
Naah, no issue whatsoever with #(true) instead command.
is this tmux CVS HEAD?
yes.
--
Beautiful is writing same markup. Internet Explorer 9 supports
standards for HTML5, CSS3, SVG 1.1, ECMAScript5, and DOM L2 L3.
Spend less time writing and rewriting code and more time
On 05/10/10 09:09, Nicholas Marriott wrote:
How do you mean? Are you saying the three processes you start once a
second never exit?
Well at least to my limited understanding, no.. at least with htop or pstree i
can't see any child processes. However, using
dstat --top-io-adv -p
tmux quickly
86 matches
Mail list logo
