removal of '{' from WORD_BREAKS

On Thu, Apr 4, 2024 at 10:38 PM Michael Paquier <michael(at)paquier(dot)xyz> wrote:
> > It kind of looks like a libedit bug, but maybe we should dig more
> > deeply. I felt itchy about 927332b95e77 removing '{' from the
> > WORD_BREAKS set, and wondered exactly how that would change readline's
> > behavior. But even if that somehow accounts for the extra backslash
> > before '{', it's not clear how it could lead to '?' and '}' also
> > getting backslashed.
>
> I don't have a clear idea, either. I also feel uneasy about
> 927332b95e77 and its change of WORD_BREAKS, but this has the smell
> of a bug from an outdated libedit version.

I too felt uneasy about that commit, for the same reason. However,
there is a justification for the change in the commit message which is
not obviously wrong, namely that ":{?name} is the only psql syntax
using the '{' sign". And in fact, SQL basically doesn't use '{' for
anything, either. We do see { showing up inside of quoted strings, for
arrays or JSON, but I would guess that the word-break characters
aren't going to affect behavior within a quoted string. So it still
seems like it should be OK? Another thing that makes me think that my
unease may be unfounded is that the matching character '}' isn't in
WORD_BREAKS either, and I would have thought that if we needed one
we'd need both.

But does anyone else have a more specific reason for thinking that
this might be a problem?

--
Robert Haas
EDB: http://www.enterprisedb.com