DAO pattern Value Objects and Data Streams

Question

Suppose we have a PersonVO (java bean) with simple attributes i.e. first and last name then the DAO is quite simple and clients of the DAO layer pass VO back and forth (really cleanly).
But it is a different story if a person has a picture. If the picture is small then we
could define a field that is simple an array of bytes i.e. byte pic[], and would work ,
but will not scale if our pic becomes too large or there are lots of persons.
How can streams be used in this case ? (without dao code leaking into the client layer) and does this break the DAO pattern ?

Answer 1

I can picture not saving the image bytes themselves
but a link to a JPG or GIF on the file system.And how would that be guarenteed on a cluster? Or even per OS?
Putting streams into a database like that can choke
performance. Can't do WHERE clauses on byte
streams.For example in PostgreSQL:
CREATE TABLE images (
  name VARCHAR(32) NOT NULL,
  image bytea NOT NULL,
  imagetype VARCHAR(4),
  CONSTRAINT image_pk PRIMARY KEY (name)
);After retrieving the image into a byte[] you can either
- directly serve it to the client (browser) over HTTP through a servlet
(don't forget to set your content type through HttpServletResponse.setContentType())
- convert the byte[] to a java.awt.image.BufferedImage using javax.imageio.ImageIO.read(byte[]), and ofcourse do everything you want (including transformations) from there.
- What ever you can think off.
Now if you're passing that VO to a JSP, the link to
the image on the server file system fits nicely into
the <img> tag and Bob's your uncle.But this will still not be guarenteed to work on a cluster, across platforms, etc.
10 % 0
java.lang.ArithmeticException: / by zero;-)