Hi, Laurenz,
On Fri, Dec 19, 2025 at 10:24 PM Laurenz Albe <[email protected]> wrote:
>
> On Fri, 2025-12-19 at 20:10 -0800, Igor Korot wrote:
> > > > > What is “clientencoding in this case?
> > > >
> > > > - if PGCLIENTENCODING is set in the environment of the client
> > > > executable, that
> > >
> > > No it is not.
> > >
> > > > - otherwise, if "client_encoding" is set on the server, that
> >
> > I just checked the postgres.conf.
> >
> > This file does not have any client_encoding.
> >
> > > > - otherwise, SQL_ASCII
> >
> > Which means that this is an encoding that will be used.
>
> You can verify that with the SQL statement "SHOW client_encoding"
> in your sample program.
I added the following code in my app:
res = PQexec( m_db, "SHOW client_encoding" );
auto value = PQgetvalue( res, 0, 0 );
PQclear( res );
and the value of the "value" variable is "UTF8".
>
> > But then I don’t understand anything.
> >
> > The code I posted above worked fine on SELECT, but INSERT failed.
> >
> > If the SQL_ASCII is the encoding used both operations should fail. Or both
> > succeeds.
> >
> > Could someone explain what happened?
>
> SQL_ASCII as client encoding means that no conversion will take place.
>
> Still, the database encoding (I suspect UTF8) will govern what can be stored
> in the database. Anything that is not valid UTF-8 will be rejected.
>
> A SELECT will never cause an error - the client will just receive data
> in UTF-8.
The exact error message is:
ERROR: invalid byte sequence for encoding UTF8: 0xdf
CONTEXT: unnamed portal parameter $1
on the INSERT.
Thank you.
>
> Yours,
> Laurenz Albe
