In Koha system (versions 3.08, 3.10, 3.12) for multi script records (e.g. titles are in Latin, in Cyrillic, in Armenian)after search, when from relevance drop down menu selecting 'Title (A-Z)' the display for non-Latin records is not in correct order. If for Latin records they are displayed in correct alphabetical order, Armenian and Cyrillic records are mixed, between the records with Armenian titles starting with the letter 'A' you can see records for which the title is starting with the letters 'M', 'H', then again 'A', then Cyrillic records are coming, and after those again records with Armenian letters. Is this a Zebra problem? Any ideas? thanks Tigran
Created attachment 19297 [details] After activating the option 'Relevance'->'Title A-Z' Armenian and Cyrillic records are mixed.
I think Zebra uses a file to know how to sort - currently we probably don't have one for multiple scripts, bug 11927 could give some hints.
Problem here is not relevance but alphabetic sorting, changing the bug title to reflect this.
This arguably could be a support ticket since not enough information has been given on Zebra configuration...
(In reply to David Cook from comment #4) > This arguably could be a support ticket since not enough information has > been given on Zebra configuration... The problem is that the sort config files are language specific and you have to decide on 'one of them'. If you have multi-language items, this is not ideal: https://git.koha-community.org/Koha-community/Koha/src/branch/master/etc/zebradb/lang_defs There are now files for ru and ua - so this could be solved by changing the Zebra config to use one of them maybe.
Is this just a CHR problem? Or does it also affect ICU? Multiple languages are always tricky...
I believe ICU as well. I think I once created a merged file for us that included Greek and other alphabets to fix some issues reported by a library using ICU.
I believe this was fixed by the addition of sort files for ru and ua: https://git.koha-community.org/Koha-community/Koha/src/branch/master/etc/zebradb/lang_defs But it might still be worth thinking about a combined file for multi-language libraries? The approach of having them all in separate files and having to decide on just one appears to be flawed.