Tom Lane wrote:
Frans <fr...@geodan.nl> writes:
We have just discovered a problem with the soundex function in
PostgreSQL 8.3.7. The problem is easy to reproduce. The following query
returns the ASCII code of the soundex representation of the Greek letter Pi:
select ascii (soundex('Î '));
In PostgreSQL 8.2.6 the result would be 0 (character null). In
PostgreSQL 8.3.7 the return value is 944, which is the UTF-16 code of
this letter.
Hm, I take it you are working in database encoding utf8?
That is correct. I should have mentioned it. It is the default encoding
we use because we often deal with non-English languages. And it is
because of multilingualism that the fuzzystrmatch functions are handy.
The
fuzzystrmatch module doesn't really work with utf8 (nor any other
multibyte encoding), because it depends on the <ctype.h> functions.
What you'll probably get when applying it to non-ascii utf8 is
an invalidly encoded string.
Well, in 8.2.6 the result for non-ASCII UTF-8 was an empty string (ASCII
code 0). You could argue that this is a valid way of expressing that the
input string could not be processed (especially if it were documented).
The nice thing about this approach is that the result is valid ASCII
(and UTF-8).
This is a known limitation that probably should be better documented.
It was just as broken in 8.2 (and every previous version), though.
But it seems there has been a recent change in the handling of non-ASCII
strings. And the result of this change is that further handling or
storing of the function output has become more difficult.
regards, tom lane
Best regards,
Frans
--
Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-bugs
