Rintze asked about locale priorities yesterday, which reminded me that
this needed to be fixed in citeproc-js. It’s now been set up, subject
to confirmation that it’s not set up wrongly. I haven’t checked any
RFCs or anything, but it all seems pretty straightforward. Here’s the
overlay hierarchy, using Austrian as an example. The last-loaded
locale node is at the top of the list:
*de-AT
*de
*""
**de-AT
**de-DE
**en-US
- = in-style locales
** = on-disk locales
Note that the “base” atop which more specific locales are loaded
differs between the disk files and the in-style locales. The former
uses en-US as the base (a failsafe in case a non-existent locale is
specified), followed by the most-common dialect of the target language
(if it differs from English). The latter uses the empty locale,
followed by the undelimited language. I don’t think this is likely to
surprise anyone, but this description just in case there’s a problem
lurking in there that I haven’t spotted.
Frank