At 6:14 AM +0900 4/24/07, Henry Nelson wrote:
On Mon, Apr 23, 2007 at 10:58:44AM -0400, Chuck Houpt wrote:
You're right. The patch should use setlocale(),
NO. Locale has nothing to do with the _weighted_ preference of
the language of a document to serve to a browser.
I agree there is no direct connection between Unix's Locale system
and HTTP's Accept-Language header. However, since LC_MESSAGES is
designed to control the language used for "informative messages", and
web pages are a kind of informational message, I think it can be
useful to use LC_MESSAGES as a default for Accept-Language.
Of course, I do understand the points made about bad translations and
interactions with ill-behaved software. Clearly the current patch is
too disruptive to Lynx's default behavior in various situations.
How about I revise the patch to be an off-by-default configuration
option? For example:
LOCALE_PREFERRED_LANGUAGE:FALSE
Setting this to true would trigger the patched code, but only when
there isn't an explicit PREFERRED_LANGUAGE setting in .lynxrc or
lynx.cfg.
Since the option would be off by default (and the option would be
commented out in the sample lynx.cfg), the current behavior of Lynx
would be unchanged.
With this configuration option, packages that want to use the feature
(like Debian and Lynxlet), can simply turn the option on in their
custom lynx.cfg. This will simplify the packages, since there will no
longer be a need for multiple language-specific config files (Debian)
or hacky patches (Lynxlet).
- Chuck
_______________________________________________
Lynx-dev mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/lynx-dev
