On Mon, Jul 06, 2009 at 05:07:15PM -0700, Jim Busser wrote:
Apologies for my ongoing slowness here. The client user interface is
controlled, among other things, by:
- the machine's (or overriding user-level) LANG and LC_ settings
This defines which language a message catalog file is
searched for before falling back to the hardwired strings
(often English).
- optionally, the provision and configuration, at the machine-
level, of a
binary message catalog which provides a translation for all of the
English program message strings (per wiki FrontendI8n)
1) this is necessary for successful translation, not optional
2) GNUmed also supports local message catalogs, not only system-
wide ones
- whichever values had been provided, by the praxis, for things like
encounter types, document types etc
That is right.
There is also in the wiki "BackendI18N" which includes "How to
provide a
translated column", "How to add a translation target (language) to
the
database" and to "allow user to select a default output language".
So it is seeming to me that the selecting / setting of a database
language *does* (or has been intended to) affect what the user sees?
That is correct.
When a user would alter the 'Encounter types" in order to provide a
"Local name", does the backend expect, and hope that these
corresponded
to whatever language the backend had already been "set" to?
Since it cannot know (due to it not "speaking" the foreign
language) it must assume as much, yes.
"Set to" being "for that database=GNUmed user".