Re: master cd88f6de4b: Adjust zero-width grapheme clusters so they are displayed (Bug#50951)

[YAMAMOTO Mitsuharu <mituharu@math.s.chiba-u.ac.jp>]

On Tue, 27 Sep 2022 11:47:25 +0900,
Po Lu wrote:
> 
> YAMAMOTO Mitsuharu <mituharu@math.s.chiba-u.ac.jp> writes:
> 
> > branch: master
> > commit cd88f6de4be1f8eba1db038b371d769f584be53b
> > Author: YAMAMOTO Mitsuharu <mituharu@math.s.chiba-u.ac.jp>
> > Commit: YAMAMOTO Mitsuharu <mituharu@math.s.chiba-u.ac.jp>
> >
> >     Adjust zero-width grapheme clusters so they are displayed (Bug#50951)
> >     
> >     * src/composite.c (composition_gstring_adjust_zero_width): New function.
> >     * src/composite.h: Declare it.
> >     * src/font.c (Ffont_shape_gstring): Use it before putting gstring to cache.
> 
> This gives:
> 
> In file included from composite.c:29:
> In function ‘AREF’,
>     inlined from ‘composition_gstring_adjust_zero_width’ at composite.c:839:11:
> lisp.h:1941:35: warning: ‘glyph’ may be used uninitialized [-Wmaybe-uninitialized]
>  1941 |   return XVECTOR (array)->contents[idx];
>       |          ~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~
> composite.c: In function ‘composition_gstring_adjust_zero_width’:
> composite.c:816:19: note: ‘glyph’ was declared here
>   816 |       Lisp_Object glyph;
>       |                   ^~~~~
> 

This is a false alarm.  The variable `glyph' is left uninitialized
only when i == LGSTRING_GLYPH_LEN (gstring), and all the uses are
guarded by this condition.  So the usual practice is to add UNINIT in
this case?

				     YAMAMOTO Mitsuharu
				mituharu@math.s.chiba-u.ac.jp