That is because the images are not stored in the table. They are stored
outside of the table in their own segment. And they have their own segment
management. But you should be able to delete the rows and add them. The
rows are allocated against a funnily named segment I bet and not the table
segment.
Jim
<jumaga_at_my-deja.com> wrote in message news:8uha7h$8ik$1_at_nnrp1.deja.com...
> I seem to have run into a very severe bug with blobs in 8i on Solaris.
> I currently have a TAR open with Oracle, and they are exploring
> officially opening a bug. I wanted to know if anyone else has had
> problems with blobs.
>
> Essentially I have a single table which contains a blob column (The
> table is in it's own tablespace.) At the beginning the table is empty,
> as is the tablespace. I then proceed to load images into the blob
> column. Everything is fine, rows are in the table, space has been
> consumed in the tablespace. Then I delete all the rows from the table,
> and then commit. A query against the table shows no rows, however the
> tablespace still shows it being full. Nothing short of dropping the
> database user frees up the space (ie. "alter table <table> deallocate
> unused" doesn't work, "alter tablespace <tablespace> coalesce" doesn't
> work, inserting an empty_blob() into the column first and then deleting
> doesn't work etc.) If you try to create an object in the tablespace
> assuming Oracle will free the space, you get an error.
>
> The thing that really has me puzzled is that the insert and then delete
> is a *fundamental* operation, not some unique situation. I'm having
> problems figuring out how such a major bug could exist.
>
> Any ideas out there?
>
> PS
> If anyone is really ambitious or curious, I have a very simple test
> case which I sent to Oracle which clearly shows the problem.
>
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
Received on Fri Nov 10 2000 - 22:24:38 CST
