Peter Samuelson writes:
>
> But that did get me thinking: if that reasoning is correct, it only
> applies to LC_CTYPE; if other locale variables (LC_MESSAGES, for
> localised output messages, LC_COLLATE, for sort order) are unusable,
> those should be mere warnings rather than errors.
>
> Thus the subversion clients would be a little more robust against
> problems with the user's environment. Does anyone else think that this
> would be a Good Thing?
>
I do. And in fact I'm wondering if this whole error out on locale error
is a false protection. Nothing stops the user from running the client with
the wrong locale, which should be as likely (or more likely?) as using a broken
ditto. What about the following scheme:
- Set LC_ALL category.
- If that fails:
- Set LC_CTYPE category
- If that fails, print a warning and continue.
I think printing a warning if setting LC_ALL is just annoying and doesn't
do much good. Warning if we can't set LC_CTYPE is good because it can explain
further strange behaviour.
Thoughts?
//Peter
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Apr 21 08:12:53 2006