On 4/3/23 18:11, Juri Linkov wrote:
>> I have now in my .emacs:
>>
>> (custom-set-variables
>>    '(tab-line-exclude-modes nil)
>>
>> but this haven't changed the behavior. The tab-line is still disappearing
>> when viewing *xxx* buffers.
>>
>> What am I still missing?
> Probably this is because you enabled buffer-local 'tab-line-mode'
> instead of 'global-tab-line-mode'.

OK, I have put

(setq global-tab-line-mode t)

into my initialization file. But this didn't help either.

Do I understand it right that the tab-line-mode is NOT as I am expecting 
and maybe erroneously assuming a property/switch of a window, but a 
property/switch of a single buffer viewed in the window??? Or is it just 
only a symbol value which might, but not must be respected while running 
rendering of the graphics of the area between the toolbar and the 
minibuffer?

And if in the buffer the local tab-line-mode is for some weird reason 
switched-off the window showing that buffer does not show the tab-line???

Coupled with not taking seriously the state of the switch shown in the 
Options -> Show/Hide -> Window Tab Line (it's just a switch showing some 
not valid switch state, like a label on a tab-bar just showing some 
non-existing buffer name as label is also considered OK and not a bug), 
this above will explain all the weird behavior I experience.

What I am wondering about is, how does it come that I am bumping in all 
these problems? Because I am trying to understand how it works and 
testing all the possible edge cases? Making around 150 windows shown in 
a "root window" sometimes called "frame" sometimes "window 
configuration" and experiencing then weird behavior on reload from the 
.desktop file, where most of the tab-lines are lost and the window sizes 
are not as before exiting and saving the .desktop file? Without being 
able to discover any clear rule or some reproducible pattern in what I 
can observe?

It seems to be not possible to create a perfectly adjusted pattern of 
100 and more windows and then store it to a .desktop file being sure 
they will be recreated on next start as seen on the screen. OK - somehow 
is it maybe misusing of an text editor for creating patterns of 
rectangles, but if the underlying programming would work the right way 
it should be possible to use it also for the unique experience of typing 
at 100 and more places on the screen at the same time.

Attached a desktop file with many, many windows which were originally 
perfectly distributed giving a nice regular pattern. The work of 
positioning the windows was not worth the effort - the restored pattern 
is all, but not as created.