Re: GCC's Unicode use [was Re: Compile performance of Linuxkernels in mainline gcc]

[Paul Schlie <schlie at comcast dot net>]

> From: "Joseph S. Myers" <joseph@codesourcery.com>
> Any programs parsing GCC's output need to set LC_ALL=C if they care at all
> about the text of error messages; this is nothing new as there have been
> translated messages for a long time.  If they only care for the message
> format "source-file:line: message" (a GNU standard, not GCC-specific) then
> they just need to expect messages in the locale character set - and
> translated messages again mean this has been the case for a long time.
> 
> There was a time when many programs adding Unicode support had their own
> special command-line options to enable it.  This was a mistake; it made
> things excessively difficult if you just want a working Unicode
> environment without fiddling with every program's configuration.  Setting
> environment variables such as LANG, LC_CTYPE and LC_ALL is the established
> way of controlling these things for all programs, not just GCC; and unlike
> five years ago, this works generally for most programs on a GNU/Linux
> system, with only a small proportion that don't yet work with Unicode.

Understood, although it would seem convenient to be able to override the
environment's local encoding preference via the command line on occasion.