Re: sed error message reports byte position instead of char position when program contains UTF-8

[Eli Zaretskii]

> Date: Thu, 16 May 2013 12:42:20 +0100 (BST)
> From: Camion SPAM <address@hidden>
> Cc: "address@hidden" <address@hidden>,
>   "address@hidden" <address@hidden>
> 
> > "Byte" is the only viable alternative, but that leaves the burden of
> > counting bytes on the user.
> 
> If you use a very long sed script, this can be a problem. 
> I happen to write sed scripts that are more than 1000 characters long.
> My last one is currently 1762 characters long and still growing.
> That's why I wrote a little bash function which would show me 
> the n'th character. but this gave wrong positions on scripts with
> UTF-8 chars. 
> 
> The work-around is to change LC_CTYPE to C around the string 
> processing part in my bash function, but, I believe that since sed 
> supports multibytes characters, the error message should count 
> characters and not bytes. btw : the error message states that the 
> position is "char" and not "byte" : 
> 
> sed: -e expression #1, char 12: unknown option to `s'

How do you expect Sed to know what character set is being used for the
command line?  Are we again going to limit ourselves to the current
locale's charset?