tint2 16.2 released :)

o9000 · 2017-05-25 18:03:49

Left-button drag is working for me. Are you saying it's not working for you? Can you post a config?

Edit: the left click must be bound to some action, otherwise it will be forwarded to the desktop.

Last edited by o9000 (2017-05-25 18:06:08)

kafran · 2017-05-25 20:21:39

Does anybody knows if there is an option like "taskbar_distribute_size" but to collapse inactive desktops? Since I have a small screen with a low resolution (1024x600), I would like to see only the taskbar of the desktop I'm in.

martix · 2017-05-25 20:30:55

This is the config I'm using (in horizontal with minor cosmetic changes). So drag&drop is supposed to work under the icons within the same desktop? Well, it only works for me if I drop it on tint2 to a different desktop. As I mentioned I used a right click, which changes the order of icons within the same desktop for me.

After having a closer look at the config I found:

mouse_left = toggle_iconify

I tried to comment it out - as before: drag&drop the icon worked only to the other desktop.

o9000 · 2017-05-25 21:48:38

Remove:

taskbar_sort_order = center

You can either sort tasks manually or by position. Not both.

o9000 · 2017-05-25 21:49:55

kafran wrote:

Does anybody knows if there is an option like "taskbar_distribute_size" but to collapse inactive desktops? Since I have a small screen with a low resolution (1024x600), I would like to see only the taskbar of the desktop I'm in.

See https://gitlab.com/o9000/tint2/blob/mas … kbar-pager

taskbar_mode = single_desktop

martix · 2017-05-25 21:54:17

@o9000 Ah, indeed, that was it! Thank you for the help and clarification!

kafran · 2017-05-25 22:00:19

o9000 wrote:

kafran wrote:
Does anybody knows if there is an option like "taskbar_distribute_size" but to collapse inactive desktops? Since I have a small screen with a low resolution (1024x600), I would like to see only the taskbar of the desktop I'm in.
See https://gitlab.com/o9000/tint2/blob/mas … kbar-pager
taskbar_mode = single_desktop

This is not what I asked. With single_desktop mode the taskbar stop acting as a pager and I need to use the keyboard to change workspace. I would like to see the desktop name on the taskbar and "on click" it expands and hide the other just like "taskbar_distribute_size" when the desktop is empty.

Last edited by kafran (2017-05-25 22:02:50)

o9000 · 2017-05-26 08:59:40

I see.

There is no pager in tint2.

You have the following options:

1. Switch desktop with the mouse wheel (if you have one).

2. Use as system tray pager, links: https://gitlab.com/o9000/tint2/wikis/Th … p-switcher (I have not used these in a while)

3. Create a pager using the executor plugin. Here is a text one (requires the packages x11-utils and xdotool, and the latest tint2 from backports):

execp = new
execp_command = xprop -root -spy _NET_CURRENT_DESKTOP | awk -W interactive '{ print "Desktop " $3 + 1 }'
execp_interval = 1
execp_has_icon = 0
execp_continuous = 1
execp_lclick_command = xdotool key Ctrl+Alt+Left
execp_rclick_command = xdotool key Ctrl+Alt+Right
execp_font_color = #ffffff 100
execp_padding = 0 0
execp_background_id = 0
execp_centered = 0
execp_icon_w = 0
execp_icon_h = 0

Assuming you use Ctrl+Alt+Left/Right in openbox to switch desktops.

Last edited by o9000 (2017-05-26 08:59:52)

kafran · 2017-05-26 10:32:29

Don't you think a good feature what I suggested: Make the taskbar collapse on innactive desktops ?

Last edited by Head_on_a_Stick (2017-05-26 22:16:02)

o9000 · 2017-05-26 12:50:43

You haven't convinced me why the other options are not good enough for you.

Anyways, even if this were useful, I do not have time to implement new features in tint2 for a while. The project will still be maintained, but there will be only bugfixes.

Patches are welcome. But they will have to be relatively complete to be merged, meaning config option handling, thorough testing with various configs and full tint2conf support.

kafran · 2017-05-28 17:48:17

o9000 wrote:

You haven't convinced me why the other options are not good enough for you.

https://www.dropbox.com/s/awtcly8qss6xt … -35-22.png

Well, I'm using tint2 on a 10.1"/1024x600 netbook. So, as you can see I have few screen space, but I like to use two desktops. If the desktop 2 could be hide and only expands when active, I think it would be a great feature. I hope someone buy the idea and implement it ^^.

Thank you for your great job with tint2.

martix · 2017-05-28 19:25:56

@kafran I had similar thoughts, but created more space by making the systray area much smaller. E.g. time and date could be put in a conky, just like the battery status -> more space for desktops. Or it would be also possible to put only the systray somewhere else (e.g. bottom right vertikal). Maybe not the best solution, but it's still an option.

o9000 · 2017-05-29 10:00:15

I see, small screen space, no mouse wheel. I will keep your request in mind.

ohnonot · 2017-05-30 15:56:27

but isn't that possible in tint2? that unused workspaces collapse, only the active one is fully expande? i'm 99.9% sure of that.

kafran · 2017-06-02 21:36:36

ohnonot wrote:

but isn't that possible in tint2? that unused workspaces collapse, only the active one is fully expande? i'm 99.9% sure of that.

For the unused (empty) workspace, yes! For the inactive one, no.

o9000 · 2017-06-11 09:20:08

0.14.6 out:

- Fixes:
- Take into account border width when computing text height (Fixes http://imgur.com/a/44xkz )
- Taskbar: Fix task icon size limits (Fixes https://forums.bunsenlabs.org/viewtopic.php?pid=51884 )
- Executor: Do not output last line if it is not terminated by newline (Fixes https://forums.bunsenlabs.org/viewtopic … 834#p52834 )
- Enhancements:
- Re-execute tint2 on SIGUSR2.
This is useful for preserving config options and environment when updating tint2. Inspired by tint3

Nili · 2017-06-11 13:12:43

Thanks so much o9000 for all efforts and fixing.

o9000 · 2017-06-20 03:15:13

kafran wrote:

o9000 wrote:
You haven't convinced me why the other options are not good enough for you.
https://www.dropbox.com/s/awtcly8qss6xt … -35-22.png
Well, I'm using tint2 on a 10.1"/1024x600 netbook. So, as you can see I have few screen space, but I like to use two desktops. If the desktop 2 could be hide and only expands when active, I think it would be a great feature. I hope someone buy the idea and implement it ^^.

Thank you for your great job with tint2.

I'm testing a new config option taskbar_hide_different_desktop, if you can compile from source it would be great to get some feedback. I don't like how the buttons change position on desktop switch, however. It would make more sense to show all desktop names together, but that is a very difficult change. Given how messy the taskbar code is, that's probably not going to happen.

So now I am thinking, keep it like this? Or revert it?

Opinions?

o9000 · 2017-08-31 19:22:29

Some info on performance metrics for tint2.

I think someone (maybe ohnonot?) asked me about how the performance of tint2 has evolved since 0.11, given that new features were added.

For this I implemented support for measuring the framerate (FPS) achieved by the panel (which I define as 1/d where d is the duration of an event loop in seconds; for example, if it takes 20 milliseconds to process events and draw the panel on screen, the framerate is 1/0.020 = 50 frames/second).

To print this for every frame, you have to set the enviroment variable DEBUG_FPS=1, i.e. run tint2 like this:

DEBUG_FPS=1 tint2

which leads to seeing a bunch of lines like this in the output:

frame 2: fps = 72 (low 38, med 38, high 80, samples 2) : processing 0%, rendering 100%, flushing 0%

I backported this to all major versions starting from 0.11.

This is used by a script that runs the panel for a few times with various configs in a virtual X server (Xvfb). The workload is quite simple, consisting for now of a script that keeps starting and stopping gnome-calculator and adding/removing/switching desktops randomly.

In addition to the minimum FPS achieved (i.e. the slowest frame), the script also reports the amount of memory used (RSS) and any memory leaks that have been detected.

The reports are here:

* master
* 0.14
* 0.13
* 0.12
* 0.11

Please keep in mind that memory use is a little noisy (if you run tint2 a few times you'd get slightly different values, since each time it might process events in a different order, leading to a different pattern of allocations/deallocations); I think up to 1 MB of variation is normal.

There is no significant difference in memory use between the 5 versions tested.

Regarding the minimum FPS, the values are *very* noisy. When running the same test multiple times, I've seen some results in tens, some in the hundreds. I think that multiple runs should be used to determine them more accurately. Anyways, here are the raw values side by side:

0.11    master (higher is better)
77.0    136.0  (test doesn't count since the config uses executors)
155.0   63.0   worse
139.0   448.0  
157.0   92.0   worse
168.0   125.0  worse
412.0   62.0   worse
193.0   127.0  worse
399.0   270.0  worse
505.0   102.0  worse
260.0   268.0
103.0   80.0   worse
204.0   178.0  worse
167.0   355.0
154.0   304.0

So for 4 out of 13 tests, the latest code is faster than 0.11; for the rest it is slower.

IMHO these results are inconsistent enough that there is probably no significant performance difference.

From black-box measurements like this it's hard to draw conclusions regarding to what might be better or worse (unless there is a clear, consistent difference that suggests a regression or an optimization).

For this reason, I've also implemented tracing support. This means that the code monitors itself and records exactly how much each operation took. Every time it sees a slow frame (under 60 frames/s), it prints the record to the terminal; otherwise it ignores it.

I've documented how it works and how the result should be interpreted here: https://gitlab.com/o9000/tint2/wikis/Tracing

I've played with it only this evening, but it already showed some interesting results. Two operations that consistently take more time (a few milliseconds) than everything else are:

* Launching an executor. Probably fork() is slow.
* Computing text size when panel elements are redrawn or resized. Unfortunately this code talks to the X server and also makes calls to the Cairo and Pango libraries, so it's not so easy to figure out what exactly is slow about it without some work -- and there is little room for optimizations.

I should point out that tracing by itself may make the code slower; and so does memory leak checking. Ironically this might be what made fork() slow in the above case. By the way, this functionality must be enabled explicitly at compile time and requires a debug build, the code is not present in packaged versions.

The DEBUG_FPS option is always enabled (but turned off by default), so you can try it if you wish.

Last edited by o9000 (2017-08-31 19:28:26)

o9000 · 2017-09-02 13:25:24

I released 15.0.

The version numbering scheme has changed, as described here https://gitlab.com/o9000/tint2/blob/15.0/ChangeLog

This release is a bit bigger than what I'd like, with bugfixes, new features, and also some tidying up of the code. Seems to work fine on my machine (TM), and there has been no negative feedback from the guys running tint2-git over at Arch, so here we go.

#121 2017-05-25 18:03:49

Re: tint2 16.2 released :)

#122 2017-05-25 20:21:39

Re: tint2 16.2 released :)

#123 2017-05-25 20:30:55

Re: tint2 16.2 released :)

#124 2017-05-25 21:48:38

Re: tint2 16.2 released :)

#125 2017-05-25 21:49:55

Re: tint2 16.2 released :)

#126 2017-05-25 21:54:17

Re: tint2 16.2 released :)

#127 2017-05-25 22:00:19

Re: tint2 16.2 released :)

#128 2017-05-26 08:59:40

Re: tint2 16.2 released :)

#129 2017-05-26 10:32:29

Re: tint2 16.2 released :)

#130 2017-05-26 12:50:43

Re: tint2 16.2 released :)

#131 2017-05-28 17:48:17

Re: tint2 16.2 released :)

#132 2017-05-28 19:25:56

Re: tint2 16.2 released :)

#133 2017-05-29 10:00:15

Re: tint2 16.2 released :)

#134 2017-05-30 15:56:27

Re: tint2 16.2 released :)

#135 2017-06-02 21:36:36

Re: tint2 16.2 released :)

#136 2017-06-11 09:20:08

Re: tint2 16.2 released :)

#137 2017-06-11 13:12:43

Re: tint2 16.2 released :)

#138 2017-06-20 03:15:13

Re: tint2 16.2 released :)

#139 2017-08-31 19:22:29

Re: tint2 16.2 released :)

#140 2017-09-02 13:25:24

Re: tint2 16.2 released :)

Board footer