On 06.03.23 11:50, I wrote:
> I guess this confusion is happening because xml_parse() was being
> called with the database encoding from GetDatabaseEncoding().
>
> I added a condition before calling xml_parse() to check if the xml
> document has a different encoding than UTF-8
>
> parse_xml_decl(xml_text2xmlChar(data), NULL, NULL, &encodingStr, NULL);
> encoding = encodingStr ? xmlChar_to_encoding(encodingStr) : PG_UTF8;
>
> doc = xml_parse(data, XMLOPTION_DOCUMENT, false, encoding, NULL);
It seems that this bug fix didn't change the output of the CI on Debian
+ Meson, 32bit.
I slightly changed the test case to a character that both encodings can
deal with.
v3 attached.