I am working with a 3rd party application that provides its own
public APIs for accessing its data. Included is an implementation of
a database ResultSet. The object provides a method for getting a
BLOB:

ByteArrayInputstream getBlob(int)

BLOBs are persisted in an Oracle 9i Database as LONG RAW types.

The problem I am encountering is related to the retrieving of a CSV
file using the code below.

ByteArrayInputStream blob = null;
...
get the blob from the db
...
csvFile = new BufferedReader(new InputStreamReader(blob));

In 3 out of 4 implementations this works fine, but in one it returns
garbage. I am told by the DBA that the 4th database is configured
identically to the others.

The content should be US-ASCII but I get mostly non-ASCII back and
actually mentions of "Adobe Photoshop CS Macintosh" in the middle of
it. This makes me think the wrong memory is being read or something.

I have tried creating the InputStreamReader with different named
charsets with no success, so I was hoping someone might have a few
more pointers as to what might be going wrong, or things to check.

Thanks in advance,

Gary