MySQL Bugs: #32101: com.mysql.jdbc.jdbc2.optional.ConnectionWrapper returns null in prepareStatement

Bug #32101	com.mysql.jdbc.jdbc2.optional.ConnectionWrapper returns null in prepareStatement
Submitted:	5 Nov 2007 12:21	Modified:	3 Dec 2007 8:21
Reporter:	Alex DeNeui
Status:	Closed
Category:	Connector/J	Severity:	S2 (Serious)
Version:	5.1.5	OS:	Any
Assigned to:		Target Version:

Description:
Three of the prepareStatement method overloads in ConnectionWrapper fail to return the
value from PreparedStatementWrapper.getInstance().  Instead, they fall through and return
the "unreachable" return null;

On line 729:

	public java.sql.PreparedStatement prepareStatement(String arg0, int arg1)
			throws SQLException {
		checkClosed();

		try {
			PreparedStatementWrapper.getInstance(this, this.mpc, this.mc
					.prepareStatement(arg0, arg1));
		} catch (SQLException sqlException) {
			checkAndFireConnectionError(sqlException);
		}

		return null; // we don't reach this code, compiler can't tell
	}

How to repeat:
1) Install 5.1.5 of Connector/J
2) Call connection.prepareStatement(sql, Statement.RETURN_GENERATED_KEYS)
3) Result is null

Suggested fix:
Add a "return"

Hello,

First of all, I think any bug report with pointing where the problem is deserves a prompt
response to the reporter. IMHO, if a SW vendor leaves a bug report alone for over 10 days
especially when it is blocker and the solution is obvious, then that vendor start to loose
community support as people value their inputs being reviewed.

Now, 

I think this bug is a major major problem that makes the connector completely unusable.

I tried to get the source, make the correction and compile it for myself.

Source does not compile as well.

I kindly ask someone from MYSQL please get down on the issue and solve it ASAP as it is a
plainer blocker for thousands of Connector-J users.

Regards,
Hasan Ceylan

just for the record:

Same problem exists in all the prepareStatement methods...

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/37954

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/37957

Thanks for the response.

The solutions is obvious. I already had a corrected version of source myself. But it does
not compile. Some interface methods are not implemented in a number of places.

I opened up a different bug for that - http://bugs.mysql.com/bug.php?id=32466 .

What's the plan for the next release? If it is not soon, what is the way to get
connector-j to copile,

Regards,

Hasan Ceylan

Because Connector/J 5.1 implements both JDBC-3 and JDBC-4 in the same driver, it can not
be compiled with JDK6 alone. You need to use JDK-1.4.2 or JDK-1.5 in your JAVA_HOME, and
point to your JDK6 javac binary, and JDK6 rt.jar using the "com.mysql.jdbc.java6.javac"
and "com.mysql.jdbc.java6.rtjar" properties, respectively.

This information is near the top of build.xml and is also in the CHANGES file and the
CHANGELOG in the online manual.

Bug#32466 has been marked as "not a bug", since the required steps are documented in files
that are pointed out in the documentation as something one should read when upgrading.

Once Connector/J has reached GA status, we generally try and release every other month or
so, depending on bug inflow. Currently there are only three bugs fixed in the 5.1 branch,
so it will probably be early December before a release is cut.

You can always download snapshot builds at
http://downloads.mysql.com/snapshots.php#connector-j

These are built nightly at 00:00 GMT.

Once again thanks for helping out Mark...

Regards,
Hasan Ceylan

One thing to clarify:

CHANGE, change.log etc files are for the *one*s who are upgrading. I am not upgrading. My
source is compliant with JDBC and and connector-J 5.1.5 and started the my project with
this version and blocked by a bug.

I *did* read the README, README.txt, doc/README doc/README.txt (which the latters are the
same files),  

The generally accepted rule is the compiler instructors (be it Makefile, build.xml, etc)
should make necessary checks for the building environment and report unmet criterias
before going ahead and trying to compile. AutoMake is superb in that, and ant still has
room for improvement.

And also sources in general include a file called BUILDING, BUILD etc. or similar file if
the information is not included in README.

Build instructions in build files (Makefile, build.xml) are for developers of the files
itselves, not for the audience.

If these are not met, like in this incident, the developer looks bad.

And the last thing, what I learned from the open source community ethic is, instead of
trying to point fingers on people, developers should try to read between the lines of
comments to find more areas to improve.

and IMHO, if the person I am dealing with is a valuable contributer and event someone who
holds 30+ paid licenses, submitting patches, then I would think twice before posting...

Regards,

Hasan Ceylan

Hasan,

I didn't consider my reply as pointing fingers. I simply stated what one needed to do, and
that in general, people should read the CHANGES file. Perhaps I made in error in saying
upgraders, since really, anyone should be aware of changes made that might affect their
application, so *everyone* should take a look at the changelog before using a piece of
open source software.

I agree that we could do better checks of the build environment. As you state, Ant doesn't
make that possible in some cases either. We'll see what we can do to fix that, but in
general the capabilities are pretty weak.

I've set this bug to "documenting", as the documentation team should make the required
components for building part of the manual.

Thanks for using MySQL.

Thanks for clarifications...

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/38019

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/38024

A note has been added to the 5.1.6 changelog: 

When using a connection from ConnectionPoolDataSource, some Connection.prepareStatement()
methods would return null instead of the prepared statement.