On Tue, Oct 14, 2003 at 04:38:12PM +0100, amck@utvinternet.com wrote:
> Why po-debconf ? gettext also supports shell, etc. (which is
> how redhat, etc. do it). Some /etc/init.d files , such as
> pcmcia, etc are shared across plaforms, which we would want,
> and they use GETTEXT. Why not just use gettext, placing the
> po files in the same place as redhat /mandrake, etc?
Yes, that's how it should be done (and makes transition to a po-based
system easier)
>
> (Obviously use po-debconf for debconf stuff, but boot and
> /etc/init.d services would use
> if [ -e /bin/gettext ]
> GETTEXT="getttext --domain $SERVICE "
> else
> GETTEXT="echo -n"
>
> $GETTEXT "Starting service..."
>
> etc.
Just to nitpick:
/etc/init.d/service should source a file (in RedHat this file is under
/etc/sysconfig/i18n IIRC) which defines:
if [ -e /bin/gettext ]
ECHO="getttext --domain $SERVICE "
else
ECHO="echo -n"
$ECHO "Starting service..."
(which is better for maintainers since you can ask them to just sed 'echo
-n' into $ECHO and they might see that as a small change).
Now, notice that some strings (echo "done." or echo "Usage: $N
{start|stop|restart|force-reload}") could be translated globally
instead on a per-domain basis.
Also notice that many of the programs called by most init scripts (such as
start-stop-daemon) should be i18n/l10n'ized or otherwise you are only
translating half of the script.
Finally, in order to be able to make this real you have to:
- make the proper implementation (the file which is sourced by init
scripts and the infraestructure for translations)
and then
a) force it through policy (unlikely)
or
b) make a mass-filing of bugs asking "pretty please" for maintainers to
change the init scripts while
One of the good side effects of having all "echo"s in init.d scripts being
under some kind of control over this information which would help solve
#169600 or even include the nice RedHat feature of colorising the output of
start-stop daemon (green OK, red NOK).
IMHO a generic mechanism for "echo"s in init.d scripts should be defined
and mandated by policy and then, an i18n implementation of this could be
sought.
Regards
Javi
Attachment:
pgpue5oqR23_R.pgp
Description: PGP signature