Neil writes all the right things:

> I wouldn't clutter Image's simplicity (and strength) with blobs, but rather
>  store the pointer to the image, and keep the actual images as native files.
>
>  One technique we have used to manage over 10 million scanned images of
>  medical claims and correspondence is to assign each image a serial number
>  (say, 10 digits), and store this in a field in an image dataset.
>
>  We store the actual images in a series of slashed sub-directories. This
>  structure could be on your HPe3000 in the Posix file space, or in our case,
>  on NT boxes, the only thing I've really found they are good for ;).

The last thing you want to do is clutter IMAGE up with inline BLOBs,
especially given that BLOBs have no definition. They can be as simple as just
a few bytes of text, or a gigabyte-long, full-length feature movie (with
separate language tracks and directors' comments).

I too believe that pointers are the only practicable way to go. I personally
have grown fond of storing standard URLs in IMAGE databases and using those
URLs for direction into an auxilliary PC operating as an FTP or HTTP server,
a technique almost identical to Neil's.

If the BLOBs are stored in-line, within IMAGE, and you have any number of
them, backups are going to become intolerably long -- as are serial searches.

It's much better to think of the data held in an IMAGE database as the
subject index that appears in the back of a textbook, equating the subject
with the page numbers on which that subject can be found, and the BLOBs as
the pages themselves. The BLOBs don't change all that often, thus I've never
been particularly worried about their security (backups can be infrequent) --
and more and more likely will stored on some other machine (most likely on
DVDs and CDs) rather than on the HP3000 itself. But the database on the
HP3000 will be searched at high speed extremely often and should provide very
fast and very direct access into the BLOBs, exactly in the manner that an
index in a book provides.

However, IF you really did want to store BLOBs in IMAGE, you can do that
right now, at least up to a file length of about 4K, which is large enough
for a signature or a small image. To do this, simply create a dataset with
perhaps only two items, a unique ID number for the record and a 4K long
dataitem. IMAGE doesn't check to see if what you write into this very long
dataitem makes any sense or not. It's just bits to IMAGE, but if your
interpreting program thinks of it as an JPEG file, then that's what it is.

Wirt Atmar