Thomas Steinmaurer wrote:

> >> I'm using FirebirdCS 2.1.2 on SuSE 11.1 Linux and I create a database
> >> specifying the default character set to be UTF8. I then connect to
> >> the database and specify that I'll be using UTF8 for the connection.
> >> If I then select on any "char" type column, I get back a string that
> >> is 4 times the length it should be and blank padded.
> >>
> >> So for instance if I have a char(1) column and store 'Y' or 'N' in it
> >> and select on it, I get back 'Y ' or 'N '. The problem doesn't
> >> happen with varchar columns and if I change the character set used for
> >> the connection to ISO8859_1, then the problem goes away.
> >>
> >> I expect that if the database is UTF8 and I'm talking UTF8 to the
> >> server and I store 'Y' in a char(1) column, when I select from it I
> >> should get 'Y' back again!
> >>
> >> PS: this is all using the C fbclient library and I'm setting the
> >> connection character set using isc_dpb_lc_ctype in my connection
> >> parameters.
> >
> > I've seen this when using IBObjects, can someone else confirm this too?
>
> Both, DBW and isql give the following result:
>
> ax
>
> when doing the following query:
>
> select c || 'x' from t2
>
> where c is a CHAR(1) in a UTF8 database. Firebird 2.1.2.
>
> HTH.

If my memory serves me right re. what the problem was in FR, it is some
struct from fbclient that specifies max data length, which FR used to
pad char fields. This was incorrect, because it's a mbcs. There's
another field in the struct that indicates the char length.

Kjell
--
--------------------------------------
Kjell Rilbe
DataDIA AB
E-post: kjell@...
Telefon: 08-761 06 55
Mobil: 0733-44 24 64