I've been subscribed to the list for about a year now, but this is my first post.
I pushed some code to https://github.com/bd-infor/pgjdbc that implements the JDBC 4.1 PreparedStatement.setCharacterStream(int, Reader) API and I am asking for feedback.
The implementation builds on work done by Alexis Meneses in pull request "Support for setBinaryStream with unknown length #220". https://github.com/pgjdbc/pgjdbc/pull/220
1) I did not add a jdbc41 counterpart to Alexis' org/postgresql/test/jdbc4/BinaryStreamTest.java. It is not clear to me if there is a character set requirement for the test database that is used by the pgjdbc tests that are run by "mvn package"? Many of the interesting test cases for Reader -> UTF-8 stream involve surrogate pairs which would require the test db to be UTF-8.
2) At present, there is no pgjdbc/src/test/java/org/postgresql/util directory. As a possible solution for item 1, I wrote a JUnit test that uses surrogate pairs directly against ReaderInputStream. It didn't seem appropriate to add it to the jdbc41 directory and I also didn't feel comfortable arbitrarily creating a new directory so I didn't push it with the other changes.
3) In addition to passing a Reader to setCharacterStream(), these changes allow you to pass a UTF-8 stream by calling PgPreparedStatement.setObject(int, Object, int, int) with Types.LONGVARCHAR. I'm testing the Types.LONGVARCHAR case with "if (in instanceof InputStream)", but it feels a bit kludgy. Any suggestions for a better way of tagging UTF-8 streams would be welcome. I considered adding a UTF8InputStream interface (or annotation), but having to subclass or wrap FileInputStream as a UTF8FileInputStream also seems kludgy.
4) setObject(int, Object, int, int) is also the one spot where I see the possibility of a backwards compatibility problem (if someone was calling setObject(..., myInputStream, LONGVARCHAR, ...) and had a working toString() that converted the stream to a String as opposed to using the default Object.toString()).
