bug#70594: 30.0.50; Error when tabbing with only one active widget

[Eli Zaretskii]

Ping!

> Cc: 70594@debbugs.gnu.org
> Date: Fri, 26 Apr 2024 18:02:13 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> 
> > Date: Fri, 26 Apr 2024 14:43:42 +0200
> > From:  Stephen Berman via "Bug reports for GNU Emacs,
> >  the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org>
> > 
> > Commit 91333dacfa1, which adds the feature of optionally skipping over
> > inactive widgets when tabbing (bug#70413), fails to handle buffers that
> > contain only one active widget; in that case, pressing TAB or S-TAB
> > raises the error "No buttons or fields found".  To reproduce:
> > 
> > 0. emacs -Q
> > 1. Load the attached file widget-test.el and then type `M-x
> >    my-widget-test'.
> > 2. In the buffer "*My Widget Test*" repeatedly press TAB and S-TAB and
> >    observe that point moves successively between the active radio
> >    buttons labelled "One" and "Two" and the active push button "Submit",
> >    but skips the inactive push button "Reset", since the user option
> >    widget-skip-inactive is enabled.
> > 3. Now press (via RET or mouse-click) the "Submit" button, which makes
> >    the radio buttons and the "Submit" button inactive and the "Reset"
> >    button active.
> > 4. Pressing TAB once moves point to the "Reset" button, then pressing
> >    TAB again moves point to BOB and raises the error "No buttons or
> >    fields found".  Likewise, pressing S-TAB once moves point to "Reset",
> >    then pressing S-TAB again moves to EOB and raises the same error.
> > 
> > The attached patched fixes this bug.  In addition, when only one widget
> > is active and point is on it, since pressing TAB or S-TAB does not move
> > point, the informational message "Only one tabable widget" is displayed.
> > 
> > The original code in widget-move contained separate loops to handle
> > forward and backward movement and this involved some code duplication.
> > The changes to this code to fix the bug would have required further
> > duplication, so I took the opportunity to parameterize the
> > directionality, allowing handling both directions in a single loop.
> > This shortens the code and IMO makes the similarities and differences in
> > the handling easier to see.
> 
> Adding Mauro to the discussion.
> 
> 
> 
>