Re: [PATCH] multibyte_identifiers: opt-in UTF-8 variable names (UAX #31 + NFC) behind a root-controlled system switch

Dear Ruben,

My two cents,

Le Mon, Aug 03, 2026 at 11:18:23PM +0000, Ruben.Vanlus via Bug reports for the GNU Bourne Again SHell a écrit :

> I can't agree with that... Today I started migrating internal company scripts...
> ... variable name (Σ=“<value>”) was rejected....

I alway considered using UTF as file, variable or function name is a bad idea.

I can't prevent my users to name her files using UTF8, but as I know inherent
problems with this, I don't use UTF-8 for programming!!

Let me show you a little sample (using bash associatives arrays):

So you use `Greek Capital Sigma' as variable name, but...

  buildUtfSample() {
      local tmpStr;
      printf -v tmpStr '["%b"]=RANDOM ' \
              \\U{3a3,1D{6BA,6f4,72e,768,7a2}}{,\\UFE0{E,F}}
      declare -gAi utf8Sample="($tmpStr)"
  }

  buildUtfSample

  for idx in "${!utf8Sample[@]}";do
      printf ' \47%s\47=%s\n' "$idx" "${utf8Sample["$idx"]}"
  done

Could produce something like:

   '𝚺️'=17943
   '𝚺︎'=13799
   '𝝨'=23400
   'Σ'=11963
   '𝚺'=11151
   '𝜮️'=5277
   '𝜮︎'=19295
   '𝝨︎'=164
   '𝝨️'=28563
   '𝞢️'=5186
   '𝞢︎'=21189
   '𝜮'=26356
   'Σ️'=8321
   'Σ︎'=1360
   '𝛴︎'=19735
   '𝛴️'=27787
   '𝞢'=12034
   '𝛴'=27951

... Then my question is: which one?

As I already said: Using UTF8 in code could
 - improve ability to obsfuscate
 - introduce new bugs
 - then new exploits... CVE-2026-44288
... and so on...

-- 
 Félix Hauri  -  <[email protected]>  -  http://www.f-hauri.ch