Doubts about Samba's unicode translation tables

[Jeremy Allison]

On Fri, Apr 19, 2024 at 11:04:40AM +0200, Xavi Hernandez via samba-technical wrote:
>Hi all,
>
>I'm currently trying to integrate Samba with CephFS, and one of the
>important things to improve is to access CephFS files in an insensitive way
>without needing to scan the entire directory from smbd.
>
>During this work I've found that Samba does the case insensitive comparison
>using a couple of UTF16 translation tables (one that converts to uppercase
>and another that converts to lowercase).
>
>Looking at how NTFS does the same thing I've found that it also uses a
>UTF16 table stored in the $UpCase special NTFS file located in the root of
>the volume.
>
>The first question is why Samba uses two tables while Windows only requires
>one ?
>For what purpose is the lowercase translation table in Samba used ?
>Is the Samba's case-insensitive comparison method actually equal to Windows
>?
>
>I've also extracted the $UpCase file from a Windows 11 machine and I've
>found that the Samba's uppercase table is very similar but not identical
>(there are 339 different values). Is this expected ?

Posting the differences would be very interesting. Thanks !