M.-A. Lemburg wrote:
Kwon, Chan Young wrote:
Hi,
Let me describe more...
1. My suggestion of LOB object is kind of Locator object. It does not contain
whole massive data. It just passes small piece of data when its methods are
called.
2. buffer object is not enough to handle various kinds of LOB types. For
example, character LOB, binary LOB and unicode character LOB.
In my company, this LOB class is already used as extension of DBAPI spec 2.0.
All colleagues like LOB class because it is very simple and easy to use.
So I sent this for your information.
There are various ways large binary/text objects are handled by
databases at the API level.
Some databases provide APIs which allow reading/writing such
data in chunks, others provide file descriptors which can be use
used for this, yet others maintain the files outside the database
and provide "locator" objects for these.
Since all of these approaches use a file-like interface in one
way or another, perhaps we could agree on a common DB-API extension
that defines the minimum interface of such objects and a constructor
that turns an existing file into an object which can then be passed
to the database (much like the Binary() constructor we have for
binary data).
Do you have documentation for the LOB object you are using available
on the web ?
Excellent idea, here are Java Docs for the JDBC clob/blob locator classes:
http://download.oracle.com/javase/1.4.2/docs/api/java/sql/Clob.html
http://download.oracle.com/javase/1.4.2/docs/api/java/sql/Blob.html
If we add this, I think we need clearly separate classes for the 2
types. In the same way Python 2.x has str/unicode and Python 3.x as
byte/str.
Chris
_______________________________________________
DB-SIG maillist - DB-SIG@python.org
http://mail.python.org/mailman/listinfo/db-sig
