Submitted by: @mrotteveel

I have a table:

create table test_clob (
id integer,
utf8_blob blob sub_type text character set utf8
)

I insert a row with UTF-8 characters using a connection with connection character set UTF8 and a value of "\u16a0\u16a1\u16a2\u16a3\u16a4\u16a5\u16a6\u16a7\u16a8\u16a9\u16aa\u16ab\u16ac\u16ad\u16ae\u16af" (where \unnnn is a unicode escape).

If I query that table using connection character set WIN1252 (or another single byte character set), the fetch will yield an op_fetch_response with code 100 (no more rows) and row count 0. Doing the same on Firebird 2.5.6 yields error "Cannot transliterate character between character sets [SQLState:22018, ISC error code:335544565]" which is more correct (although still very annoying, I would rather have that error when I fetch the blob).

Reproducing with isql (slightly reduced):

set names utf8;
connect ....;
insert into test_clob(id, utf8_blob) values (1, x'e19aa0'); -- e19aa0 is UTF-8 encoding of code point 16a0
select * from test_clob; -- yields row
set names win1252;
connect ....;
select * from test_clob; -- yields no row, instead of expected error "Cannot transliterate character between character sets"