bug#73752: 29.4; Ligatures are randomly rendered with extra spaces

[Visuwesh]

[சனி நவம்பர் 02, 2024] Eli Zaretskii wrote:

>> From: Visuwesh <visuweshm@gmail.com>
>> Cc: dev@real-or-random.org,  xuan@xlk.me,  73752@debbugs.gnu.org
>> Date: Sat, 02 Nov 2024 18:59:03 +0530
>> 
>> [சனி நவம்பர் 02, 2024] Eli Zaretskii wrote:
>> 
>> >> But just loading a file that contains
>> >> 
>> >>     (set-face-attribute 'default nil
>> >>                 :family "Cascadia Code"
>> >>                 :height 110
>> >>                 :weight 'bold)
>> >> 
>> >>     (load "/tmp/ligature.el")
>> >> 
>> >>     (ligature-set-ligatures 'prog-mode '("--" "---" "==" "===" "!=" "!==" 
>> >> "=!="
>> >>                                          "=:=" "=/=" "<=" ">=" "&&" "&&&" 
>> >> "&=" "++" "+++" "***" ";;" "!!"
>> >>                                          "??" "???" "?:" "?." "?=" "<:" 
>> >> ":<" ":>" ">:" "<:<" "<>" "<<<" ">>>"
>> >>                                          "<<" ">>" "||" "-|" "_|_" "|-" 
>> >> "||-" "|=" "||=" "##" "###" "####"
>> >>                                          "#{" "#[" "]#" "#(" "#?" "#_" 
>> >> "#_(" "#:" "#!" "#=" "^=" "<$>" "<$"
>> >>                                          "$>" "<+>" "<+" "+>" "<*>" "<*" 
>> >> "*>" "</" "</>" "/>" "<!--" "<#--"
>> >>                                          "-->" "->" "->>" "<<-" "<-" 
>> >> "<=<" "=<<" "<<=" "<==" "<=>" "<==>"
>> >>                                          "==>" "=>" "=>>" ">=>" ">>=" 
>> >> ">>-" ">-" "-<" "-<<" ">->" "<-<" "<-|"
>> >>                                          "<=|" "|=>" "|->" "<->" "<~~" 
>> >> "<~" "<~>" "~~" "~~>" "~>" "~-" "-~"
>> >>                                          "~@" "[||]" "|]" "[|" "|}" "{|" 
>> >> "[<" ">]" "|>" "<|" "||>" "<||"
>> >>                                          "|||>" "<|||" "<|>" "..." ".." 
>> >> ".=" "..<" ".?" "::" ":::" ":=" "::="
>> >>                                          ":?" ":?>" "//" "///" "/*" "*/" 
>> >> "/=" "//=" "/==" "@_" "__" "???"
>> >>                                          "<:<" ";;;"))
>> >>     (global-ligature-mode t)
>> >> 
>> >> was enough to trigger the breakpoint for me.  This is, of course, before
>> >> I got a chance to even reproduce the misalignment.
>> >> 
>> >> Here's what the gdb buffer looks like:
>> >> 
>> >>     (gdb) source .gdbinit
>> >>     SIGINT is used by the debugger.
>> >>     Are you sure you want to change it? (y or n) [answered Y; input not 
>> >> from terminal]
>> >>     DISPLAY = :0.0
>> >>     TERM = dumb
>> >>     Breakpoint 1 at 0x20b4b7: file emacs.c, line 432.
>> >>     Breakpoint 2 at 0x1cfa8a: file xterm.c, line 27102.
>> >>     (gdb) break hbfont.c:598 if xoff == 0 && yoff == 0 && wadjust == 
>> >> metrics.width && LGLYPH_ADJUSTMENT(lglyph) != Qnil
>> >>     Breakpoint 3 at 0x3c3ad7: file hbfont.c, line 598.
>> >>     (gdb) run -Q
>> >>     Starting program: /home/viz/lib/ports/emacs/src/emacs -Q
>> >>     [Thread debugging using libthread_db enabled]
>> >>     Using host libthread_db library 
>> >> "/lib/x86_64-linux-gnu/libthread_db.so.1".
>> >>     [New Thread 0x7ffff1b8f6c0 (LWP 198125)]
>> >>     [New Thread 0x7ffff118f6c0 (LWP 198126)]
>> >>     [New Thread 0x7fffebf8f6c0 (LWP 198127)]
>> >>     [New Thread 0x7fffeb58f6c0 (LWP 198128)]
>> >>     Error in testing condition for breakpoint 3:
>> >>     Attempt to take address of value not located in memory.
>> >> 
>> >>     Thread 1 "emacs" hit Breakpoint 3, hbfont_shape 
>> >> (lgstring=XIL(0x7ffff2a9c745), direction=XIL(0x33c0)) at hbfont.c:598
>> >>     598         if (xoff || yoff || wadjust != metrics.width)
>> >
>> > Good.  Now please do this:
>> >
>> >   (gdb) pp lgstring
>> >   (gdb) p wadjust
>> >   (gdb) p metrics.width
>> >   (gdb) bt
>> >
>> > And post the results here.
>> 
>>     (gdb) pp lgstring
>>     [[#<font-object "-SAJA-Cascadia
>> Code-bold-normal-normal-*-15-*-*-*-m-0-iso10646-1"> 59 59] nil [0 0
>> 59 1935 9 4 14 8 4 nil] [1 1 59 1865 9 2 6 8 4 nil] nil nil nil nil
>> nil nil]
>>     (gdb) p wadjust
>>     $1 = 9
>>     (gdb) p metrics.width
>>     $2 = 9
>
> Thanks, but this is a false alarm: the lgstring's glyphs don't have
> the [XOFF YOFF WADJUST] component.  So either my breakpoint condition
> is somehow wrong, or you mistyped it, or something else.

I don't think I mistyped the condition at least.  Here's what I typed:

    (gdb) break hbfont.c:598 if xoff == 0 && yoff == 0 && wadjust == 
metrics.width && LGLYPH_ADJUSTMENT(lglyph) != Qnil

>  What does GDB show if you type
>
>   (gdb) p LGLYPH_ADJUSTMENT(lglyph)
>   (gdb) p Qnil

    (gdb) p LGLYPH_ADJUSTMENT(lglyph)
    $3 = XIL(0)
    (gdb) p Qnil
    $4 = XIL(0)