Changing character sets in text fields

[Ben Rubinstein]

I built a Valentina database on the Mac, which included text fields into
which I carefully put text constructed in ISO-8859-1, because that was how
the application that would be using the database on Windows would be
treating it.

However, the data came out wrong on Windows - it looked just like I'd expect
ISO-8859-1 to look like on the Mac.  I then rebuilt the database (again on
the Mac) with text in the MacRoman character set, and it came out 'right' on
Windows.

Obviously Valentina had done some conversion on the data.  It's very kind of
Valentina, but it was unexpected, and it may not be what I want.

I've not been able to find any reference to this behaviour in the
documentation.  I'd like to know:

* how and where this happens.  EG, is the data stored internally in unicode,
and transformed on both storage and retrieval according to the local
platform?  Or is it always stored in one character set, and Valentina
transforms on storage and retrieval on some platforms?  Or is it always
stored exactly as inserted, with a flag to cause transformation on retrieval
in some situations?  Or some other arrangement?

* under what circumstances this is applied.  EG is it any 'text' field?
what about 'varchar'?

* can this be configured.  Is there a setting either in Valentina, or per
database or per table, that can be used to avoid this behaviour when it is
not desired?
  
TIA,
 
  Ben Rubinstein               |  Email: benr_mc at cogapp.com
  Cognitive Applications Ltd   |  Phone: +44 (0)1273-821600
  http://www.cogapp.com        |  Fax  : +44 (0)1273-728866