It would be nice, but it is not true. You can get correct non utf8 xml with encoding specification only when binary mode is used. Psql doesn't support binary mode.
Regards
Pavel
Dne 15. 4. 2015 22:06 napsal uživatel "Peter Eisentraut" <
peter_e@gmx.net>:
On 4/10/15 5:26 PM, Pavel Stehule wrote:
> Hi
>
> I wrote a prototype of this patch, and it works well
>
> postgres=# set client_encoding to 'latin2';
> SET
> Time: 1.488 ms
> postgres=# \copy (select xmlelement(name xx, d) from d) to ~/d.xml
> (format 'raw')
> COPY 1
> Time: 1.108 ms
> postgres=# copy (select xmlelement(name xx, d) from d) to stdout (format
> 'raw') ;
> <?xml version="1.0" encoding="LATIN2"?><xx>příliš žluťoučký
> kůň</xx>Time: 1.000 ms
I think you can get the same thing using regular psql output and just
turning off all field and record separators and tuple headers and so on.
