Zebra doesn't like NSB/NSE chars, they should be deleted before indexing, only in UNIMARC because it's a french problem.
(In reply to comment #0) > Zebra doesn't like NSB/NSE chars, they should be deleted before indexing, only > in UNIMARC because it's a french problem. It would be better if string contained between \x88 and \x89 was skipped when indexing (in sort charmap).
This was pushed to 3.0.x
fridolyn, we also have something in sort charmap. Except it doesn't work well if you've icu set, so this patch is relevant I think.