Thus the problem is that " .... collations are implemented using the operating system charset and locale support ... " while, other engines, implements collations internally ..... is it right ?
That's my understanding, but I don't know which other database systems you're talking about because you've never specifically named any.
It's entirely possible that some other DBMSs use the system locale and collation support with internal downcasing, for example.
All I know, I've already said, but I'm not an expert on Pg's innards.
-- Craig Ringer
Tahnk you for replay. At this point, the solution could be a new, custom, operating system collation .... (something like: en_CI_US.UTF-8) ....