On 2024-04-09 23:49, Aurelien Jarno wrote:
Hi,
On 2024-03-15 16:13, M. Buecher wrote:
Adding / back to end of the path, so that a locale file is immediately
found.
diff --git a/debian/local/usr_sbin/locale-gen b/debian/local/usr_sbin/locale-gen
index 7fa3d772..1711a4f0 100755
--- a/debian/local/usr_sbin/locale-gen
+++ b/debian/local/usr_sbin/locale-gen
@@ -23,6 +23,12 @@ is_entry_ok() {
fi
}
+if [ -z "${I18NPATH:-}" ]; then
+ if [ -d "${USER_LOCALES}" ]; then
+ I18NPATH="${USER_LOCALES%%/locales*}/"
+ fi
+fi
+
echo "Generating locales (this might take a while)..."
while read -r locale charset; do
if [ -z "$locale" ] || [ "${locale#\#}" != "$locale" ]; then continue;
fi
@@ -46,7 +52,7 @@ while read -r locale charset; do
input="$USER_LOCALES/$input"
fi
fi
- localedef -i "$input" -c -f "$charset" -A /usr/share/locale/locale.alias
"$locale" || :
+ I18NPATH="${I18NPATH}" localedef -i "$input" -c -f "$charset" -A
/usr/share/locale/locale.alias "$locale" || :
echo " done"
done < "$LOCALEGEN"
echo "Generation complete."
Thanks for your bug report and for even proposing a patch which looks
good to me.
That said as I18NPATH could actually contains a list of directory, I
wonder if the behaviour would be more consistent if the user locales
directory is always prepended to I18NPATH if it exists. What do you
think?
Regards
Aurelien
My thought was: if someone already specified I18NPATH explicitly, then
he knows what he is doing and the script should not mess with his setup.
Maddes
