bug#69305: outline-minor-mode for tabulated-list-mode

[Eli Zaretskii <eliz@gnu.org>]

> From: Juri Linkov <juri@linkov.net>
> Cc: 69305@debbugs.gnu.org
> Date: Fri, 23 Feb 2024 09:09:03 +0200
> 
> >> >> +@defvar tabulated-list-groups
> >> >> +This buffer-local variable specifies the groups of entries displayed in
> >> >> +the Tabulated List buffer.  Its value should be either a list, or a
> >> >> +function.
> >> >> +
> >> >> +If the value is a list, each list element corresponds to one group, and
> >> >> +should have the form @w{@code{(@var{group-name} @var{entries})}}, where
> >> >> +@var{group-name} is a string inserted before all group entries, and
> >> >> +@var{entries} have the same format as @code{tabulated-list-entries}
> >> >> +(see above).
> >> >> +
> >> >> +Otherwise, the value should be a function which returns a list of the
> >> >> +above form when called with no arguments.
> >> >> +@end defvar
> >> >
> >> > I think the way to specify ENTRIES for this customization should be
> >> > described in more detail.  Reading the above description, even the
> >> > idea of using ENTRIES for specifying grouping is unclear, since
> >> > tabulated-list-entries is basically just a list of column descriptors.
> >>
> >> But ENTRIES are described in 'tabulated-list-entries'.  Here ENTRIES
> >> are no different from 'tabulated-list-entries' that are just
> >> column descriptors.
> >
> > My point is that it isn't obvious how to describe grouping in terms of
> > column descriptors.  I think the manual should tell in more detail how
> > to use column descriptors to specify how entries should be grouped.
> > Maybe an example or two is all that it takes to do that.
> 
> Sorry, I don't understand what is unclear here:
> 
>   If the value is a list, each list element corresponds to one group, and
>   should have the form @w{@code{(@var{group-name} @var{entries})}}, where
>   @var{group-name} is a string inserted before all group entries, and
>   @var{entries} have the same format as @code{tabulated-list-entries}
>   (see above).
> 
> Basically this says that the format is (group-name entries) where
> entries are described in tabulated-list-entries as having the format
> (id contents).  There is no need to duplicate the description of entries.

Maybe I'm missing something here.  To put it more concretely, can you
show a value of tabulated-list-groups that will cause the buffers in
buffer list grouped by major mode using the above form?

> >> >> +(ert-deftest tabulated-list-groups ()
> >> >> +  (with-temp-buffer
> >> >> +    (tabulated-list-mode)
> >> >> +    (setq tabulated-list-groups
> >> >> +          (reverse
> >> >> +           (seq-group-by (lambda (b) (concat "* " (aref (cadr b) 3)))
> >> >> +                         tabulated-list--test-entries)))
> >> >> +    (setq tabulated-list-format tabulated-list--test-format)
> >> >> +    (setq tabulated-list-padding 7)
> >> >> +    (tabulated-list-init-header)
> >> >> +    (tabulated-list-print)
> >> >
> >> > This seems to test only the function value of tabulated-list-groups?
> >> > What about the other form of the value?
> >>
> >> The other forms have no functional difference.
> >
> > Sorry, I don't follow.  Shouldn't we test the capability of specifying
> > grouping by the method other than by providing a function?
> 
> Actually there are no other methods.

The doc string says the "value can be either a list or a function".  I
see only one value here, so I'm asking what about a value of the other
form.  Or what am I missing here?