[PATCH stable v2] curses: do not initialize LC_ALL to user settings (issue6358)

[Yuya Nishihara]

On Fri, 26 Jun 2020 14:42:58 +0200, Manuel Jacob wrote:
> On 2020-06-26 13:47, Yuya Nishihara wrote:
> > On Fri, 26 Jun 2020 10:35:24 +0200, Manuel Jacob wrote:
> >> # HG changeset patch
> >> # User Manuel Jacob <me at manueljacob.de>
> >> # Date 1593157054 -7200
> >> #      Fri Jun 26 09:37:34 2020 +0200
> >> # Branch stable
> >> # Node ID 36dcc87c0c15d8757ef0322c820861b1d10350c1
> >> # Parent  9bb2fc4ac22b0b986d56fd13be4e6b524af7b648
> >> # EXP-Topic init_locale
> >> curses: do not initialize LC_ALL to user settings (issue6358)
> > 
> > Queued these to default, thanks. The issue to be fixed is relatively 
> > small,
> > but the setlocale() change might have unwanted side effect.
> 
> As mentioned in the first patch, non-ASCII filenames in 
> hgext.convert.subversion are fixed by it. If I make further non-ASCII 
> fixes to hgext.convert.subversion, should do they go to stable (for 
> testing, I can ensure that LC_CTYPE is initialized otherwise on Python 
> 2) or default? For me as a user, it doesn’t make a difference, as I 
> don’t use Subversion (especially not with non-ASCII filenames).

If the scope of the change is narrowed down to the conversion extension,
I think it may go stable. As a user, I never use non-ASCII filenames, too.