Dirk Olmes wrote:
> Apart from the fact that the test doesn't test queries using ORDER BY as
> its name would suggest,
You are right, this is dumb :-). This is a simple LIKE test, has nothing
to do with ordering. I just committed a fix for that.
my current assumption is that the test runs on
> oracle only through the existence of the OracleQualifierTranslator.
> It seems that Oracle and Postgres share the same behaviour of CHAR(xx)
> columns: padding the column to its specified length with blanks.
> OracleQualifierTranslator seems to generate queries with RTRIM(column)
> so that the test will finally find something. Since I don't have such a
> QualifierTranslator installed for postgres, things go wrong.
I suspect this is the case. BTW, I just did some research with McKoi
database, and out of all SelectQuery tests, "testOrderByIgnoreCase" was
the only one that failed, and for the same reason - space padding.
> Can someone please explain the design goal behind the whole trimming of
> CHAR columns, please? I remember that in the past CHAR columns were
> handled more efficiently by DBMSs than VARCHAR but I remember having
> read in the Oracle docs that this is no longer true.
That's what I heard in the past too. Not sure what is the current
situation. In any event I suspect it to be database-dependent. Also
since there is a CHAR type in JDBC, we must support it correctly. It is
up to the users to avoid using it in the design.
I guess for now we may need a PostgresQualifierTranslator, just like
Oracle's. Later we may refactor it into an adapter-independent subclass
of QualifierTranslator.
Andrus
This archive was generated by hypermail 2.0.0 : Mon Feb 17 2003 - 00:29:09 EST