svn commit: r281129 - head/etc

[Stanislav Sedov]

> On Apr 10, 2015, at 12:32 AM, Gleb Smirnoff <glebius at FreeBSD.org> wrote:
> 
> On Fri, Apr 10, 2015 at 10:08:45AM +0300, Slawa Olhovchenkov wrote:
> S> What benefits from switching to utf8 locale in 'default' login class?
> 
> Being on the same page as rest of the world?
> 

To be fair, this is not exactly true.  To my knowledge no mainstream
operating systems except Linux employ UTF-8 at the moment:
* Windows is still UTF-16 based
* Mac OS X uses it’s own version of UTF-8 (aka UTF-8-MAC)
  which is the cause of constant pain when moving files between
  OS X and FreeBSD.

Personally, I’d agree with Slawa that this change seems somewhat
premature and does not seem to bring any immediate benefits:
* It will hardly improve compatibility with other OSes;
* A lot of applications in widespread use do not have UTF-8 support,
  and UTF generally requires some non-trivial handling unlike
  8 bit encodings which lead to numerous security issues in the
  past (and continues to so).
* Changing the default encoding for the login class will lead
  to an unexpected encoding change for ‘russian’ login class
  users.  This might affect a lot of legacy applications, and
  might indeed result in a data loss in case of databases.

In my view, adding an additional login class with UTF-8 encoding
sounds like a better solution at this time.  This will bring
pretty much the same benefits (as you noted earlier fresh FreeBSD
installation require specifying the login class manually, in
which case that new login class can be passed in), but at the same
time it won’t result in a sudden encoding change for existing
‘russian’ login class users. 

--
ST4096-RIPE