Re: Fix width tabs

[Eli Zaretskii]

> From: Juri Linkov <juri@linkov.net>
> Cc: emacs-devel@gnu.org
> Date: Fri, 04 Nov 2022 09:48:58 +0200
> 
> >> Firefox choose to change the foreground color on tabs end to make names
> >> easier to read.  I guess it isn't supported by Emacs. So I use to set
> >> tab-bar-tab-name-function to tab-bar-tab-name-truncated and
> >> tab-bar-tab-name-ellipsis to t.  It helps distinguish the separation
> >> between tab names.  Is it expected for the truncation to not applying
> >> anymore with the code you sent?
> >
> > But using a faded foreground color like in Firefox is
> > the right thing to do.  After adding this to the end
> > produces such nice look where tab names are fading to the right:
> 
> This feature is pushed now enabled by default.

Thanks.  However, I have problems with the documentation of this
feature:

  (defcustom tab-bar-fixed-width t
    "Automatically resize tabs on the tab bar to the fixed width.
  This variable is intended to solve two problems.  When switching buffers
  on the current tab, the tab changes its name to buffer names of
  various lengths, thus resizing the tab and shifting the tab positions
  on the tab bar.  But with the fixed width, the size of the tab name
  doesn't change when the tab name changes, thus keeping the fixed
  tab bar layout.  The second problem solved by this variable is to prevent
  wrapping the long tab bar to the second line, thus keeping the height of
  the tab bar always fixed to one line.

  The maximum tab width is defined by the variable `tab-bar-fixed-width-max'."

This begs the question: what is the fixed width to which the tabs are
resized? if it's "fixed", then the value is known in advance, right?
Moreover, if the tab sizes are fixed, why does the doc string say
"automatically resize"? "resizing" is the antithesis of "fixed width".
Can you explain in plain words what this option does, when it is
non-nil?  (There's a hint to that in the doc string of
tab-bar-fixed-width-max, but that's not the right place for this
information.)  Also, what does it do when it's nil?

The style of the doc string is also problematic: we shouldn't describe
in a doc string of user option what problems it solves.  Instead, we
should tell what is the behavior for each valid value of the option;
the judgment of what is a "problem" and what isn't is left to the
user, because the needs of users may differ, and what is a "problem"
for some is a "solution" to others.  That's why we have user options
to begin with.

The NEWS entry is also problematic, basically for the same reason.

We should fix this.