Re: GNUlib unicode encoding causes smart quotes to be displayed in program's output

[Pádraig Brady]

On 07/12/2019 04:10, Bruno Haible wrote:
> Wes Hurd wrote:
> > What I meant about smart quotes being dangerous was, if copying the output
> > text that contains smart quotes to use somewhere else (especially in code),
> > the smart quotes have to be manually replaced which is tedious for the user
> > (programmer).
>
> It's quite the opposite: The smart quote characters are safer because they
> don't interfere with the quoting of a shell or programming language. For
> example, in my music directory, I have a directory T’pau.
>
> $ ls T’pau
> China in your hand.wav
>
> If the directory was named T'pau and I were to write the command
>
> $ ls T'pau
>
> it would hang. And
>
> $ ls 'T'pau'
>
> would hang as well. *This* is dangerous.
>
> > The user may not even see that smart quotes are being used unless there is
> > a breaking error.
>
> But the smart quotes look different than an apostrophe! No confusion is 
> possible.
> Your argument was correct 40 years ago, when
>    1. the shape of the apostrophe was given by US-ASCII and was not vertical 
> [1],
>    2. a character cell had so few pixels in height and width that smart quotes
>       and apostrophes could not be rendered differently.
>
> The reason why quotearg.c uses smart quotes is explained in [2].
>
> Bruno
>
> [1] https://www.cl.cam.ac.uk/~mgk25/ucs/apostrophe.html
> [2] https://www.cl.cam.ac.uk/~mgk25/ucs/quotes.html

Agreeing with Bruno here.

Note gnulib supports shell quoting, which coreutils uses to safely quote file 
names:
https://git.sv.gnu.org/gitweb/?p=coreutils.git;a=commitdiff;h=v8.24-71-g08e8fd7
Also coreutils ls leverages this to safely output file names by default.

Note also re apostrophe, word boundaries may be a consideration.
As I previously tweeted:

  It's awkward for file names to use shell quote
  It’s awkward for word regex to use right quote (\u2019)
  Itʼs best to use apostrophe modifier (\u02BC)

cheers,
Pádraig