On Friday, July 24, 2015 4:26 AM, Steven Folsom wrote:

> For example, in my last email I posed the possibility of adding a bf:nonSort
> property for what has been handled in MARC with non-filing characters. It's
> unclear (at least to me) how long we need to bother with recording this
> information. There are sophisticated methods for building index sort values,
> but different languages seem to present different challenges. Not to
> mention, you could get a children's book titled and about the letter "A".

Yes, sorting is hard. The best solution would be that data publishers do their best to supply language tags for all strings and then consuming systems can use that information to index the data according to the conventions for that specific language. I realise that it can be tough to get correct language information out of some systems, though.