bug#58168: string-lessp glitches and inconsistencies

[Eli Zaretskii]

> From: Mattias Engdegård <mattias.engdegard@gmail.com>
> Date: Mon, 3 Oct 2022 21:48:10 +0200
> Cc: 58168@debbugs.gnu.org,
>  Eli Zaretskii <eliz@gnu.org>
> 
> 1 okt. 2022 kl. 15.51 skrev Eli Zaretskii <eliz@gnu.org>:
> 
> > I think the variable is a misnomer of sorts: the request back when it
> > was introduced to display hex where we usually display octal \nnn
> > escapes.  And the latter happens not only for raw bytes.
> 
> Fair enough. Maybe the documentation should reflect that, but I'm still 
> holding out for a change to the C1 presentation in the long term, so...
> 
> I'm not going to pursue this little digression any further except that while 
> looking at it I found a few inaccuracies and a likely bug in 
> redisplay-testsuite.el. I'm attaching a patch which un-muddles the test and 
> adds a display of unprintable Unicode chars such as C1 controls, in addition 
> to raw bytes. I'd like to adorn the commit with the correct bug number so if 
> you remember that of the original discussion that would be useful (I never 
> found it very easy to search debbugs).

First I needed to fix fallout from making STRING_CHAR intolerant of
unibyte text, because redisplay-testsuite caused assertion violations
in string_char_and_length.  (Doesn't it abort for you? or do you not
build Emacs with --enable-checking?)  This was a regression in Emacs
28, sigh.

Looking at your patch, I don't think I understand this part:

> --- a/test/manual/redisplay-testsuite.el
> +++ b/test/manual/redisplay-testsuite.el
> @@ -305,7 +305,7 @@ test-redisplay-5-toggle
>    (let ((label (if display-raw-bytes-as-hex "\\x80" "\\200")))
>      (overlay-put test-redisplay-5a-expected-overlay 'display
>                   (propertize label 'face 'escape-glyph)))
> -  (let ((label (if display-raw-bytes-as-hex "\\x3fffc" "\\777774")))
> +  (let ((label (if display-raw-bytes-as-hex "\\xfc" "\\374")))
>      (overlay-put test-redisplay-5b-expected-overlay 'display
>                   (propertize label 'face 'escape-glyph))))
>  
> @@ -320,18 +320,36 @@ test-redisplay-5
>          (test-insert-overlay " " 'display "\200"))
>    (insert "\n\n")
>    (insert "  Expected: ")
> -  ;; This tests a large codepoint, to make sure the internal buffer we
> -  ;; use to produce the representation is large enough.
> -  (aset printable-chars #x3fffc nil)
>    (setq test-redisplay-5b-expected-overlay
>          (test-insert-overlay " " 'display
> -                             (propertize "\\777774" 'face 'escape-glyph)))
> +                             (propertize "\\374" 'face 'escape-glyph)))

I could understand why you'd want to _add_ the larger values, but why
replace?

As for the bug report which led to display-raw-bytes-as-hex (if that
what you meant) and its discussion, it's bug#27122.  (If you know what
code is in question, it is much easier to find the bug via "git
annotate", assuming the bug number was cited in the commit logs.)