Thanks Nikolay/Matt.
I won't bother copying the data now that you've convinced me that it's efficient
Also I noticed the image speed issue when I was dealing with internal images.
Doing a scan of my 3000 tracks takes 3 seconds normally ... when I ask for internal images (where there is only 1 out of the 3000) it takes around 5 minutes (that's why it's disabled at the moment). Is there any way to speed this up, e.g. hold a flag in your DB if the file contains an image, the byte offset within the file that the image starts, the length of the image and the image type.
You take the hit up front (slower import) but the benefits are tremendous.
For example I could open the file as a memory mapped file and extract/paint the image onto the screen as necessary.
Also I'll do all my communications from now on thru this board ... but is it still ok to announce alpha's/beta's in the main board?
Rhino.