bug#24772: 26.0.50; homoglyph face is used in Gnus, eww

[Óscar Fuentes]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Óscar Fuentes <ofv@wanadoo.es>
>> Date: Sun, 23 Oct 2016 16:28:17 +0200
>> 
>> 
>> The new homoglyph face introduced in
>> c06a518b5b0706509252abb03875867b10a92fc7 is being used for displaying
>> characters in Gnu and eww buffers. Those characters look like sore
>> thumbs, very displeasing.
>
> Could you please mention a couple of Web pages where you see this?

It happens with this, for instance:

http://debbugs.gnu.org/cgi/bugreport.cgi?bug=24772

> Also, do you see something different in 25.1.50?

Right now emacs-25 fails to compile with

Processing OKURI-NASI entries ...
processed 10% ...
processed 20% ...
Args out of range: #("ね /黒" 0 4 (charset japanese-jisx0208)), -5, nil
Makefile:142: recipe for target 
'../../emacs/leim/../lisp/leim/ja-dic/ja-dic.el' failed
make[2]: *** [../../emacs/leim/../lisp/leim/ja-dic/ja-dic.el] Error 255
make[2]: Leaving directory '/home/oscar/dev/emacs/build/leim'
Makefile:320: recipe for target 'leim' failed
make[1]: *** [leim] Error 2
make[1]: Leaving directory '/home/oscar/dev/emacs/build/lisp'
Makefile:385: recipe for target 'lisp' failed
make: *** [lisp] Error 2


I can seek a good revision if that's important. But while trying to
reproduce the bug, I noticed that emacs -Q is not affected, nor a
regular emacs session (that loads my config) but the session I'm running
as a daemon does show the problem. I think that that is the real
trigger. Can you try visiting the web page above with a daemonized emacs
on GNU/Linux? Or can you tell me what data to inspect to know if Emacs
decided to show those chars on a translated form when the daemon
started?