Re: Extent of localisation customisation



Today at 14:56, Alan Cox wrote:

> I was anticipating a pull down of the standard formats people would
> be using not arbitary reformatting. The latter doesn't actually work
> out with or without locale because programs make assumptions about
> printf/sscanf etc.

Yes, I went to extremes in an attempt to emphasize my point: we have
to know what locales provide what formats on what systems, if all
systems even provide all needed locales.

> Providing we are offering pull downs of the common formats the user
> should be happy (unless they are very weird or we missed a format and
> need to add it). Asking the user to select between
> mm/dd/yy yyyy/mm/dd dd/mm/yy is the level I mean here

Yes, I understand, but how do we implement that?  LC_TIME is used
only to specify which *locale* to use, not which format to use.

There're a couple issues here:
- we don't know which locale contains the desired format (i.e. on
  some system, en_US might use mm/dd/yy, yet on other mm/dd/yyyy)
- we don't know even if we have locales with all the needed formats
  on all supported systems (it's not uncommon to have systems with
  only C/POSIX and single other locale installed)

So, how do we solve that?  I mean technically (and with current
systems and locale support), and what would work in practice.

(Of course, note that these three are not enough for all Gnome users;
some would expect other delimiters such as "." instead of "/", so
you'd need to find fairly stable 10-or-so locales on every supported
system; or simply go for GNU libc as The Platform, like I suggested
earlier.)

Cheers,
Danilo



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]