Clob will now use the connection encoding

[andrewtaylor]

Suggested fix to make CLOB handling with the large object store a little better. Uses the encoding from the connection to determine how the character stream should be encoded before storing it in the large object store. Not a perfect solution, but the best we can do without storing the encoding with the large object. Should at least be better than mangling the data as US-ASCII.

[davecramer]

If memory serves the connection encoding is always UTF8 with the JDBC driver.

This PR suggests that there is a bug but it might be simpler to just set the encoding to UTF-8 ?

[andrewtaylor]

Hi,

The underlying issue is on line 3177 of AbstractJdbc2Statement.java. It is calling getAsciiStream(), which by the definition of the API method can only return US-ASCII. The change I propose is that it call getCharacterStream() (which gives us a Writer) and write the data to the large object store using the connection encoding.

The change to AbstractJdbc2Clob is to ensure that the reader also uses the connection encoding rather than the current behaviour, which is to use the platform encoding.

FYI database and platform are both UTF-8, I assume the connection will be the same (I have never had to set it explicitly). All other strings are handled correctly.