TODO Notes

27 June 2007
------------
Due to the design of the metadata import process which identifies images by
their filename (with extension removed), there is a requirement to have a
UNIQUE constraint on the image.image_filename. This is potentially a
significant limitation, where all images must have a unique file name and can
cause issues especially with camera-generated filenames where the chance of a
collision would be particularly high.

A solution to this could be to introduce automatic file naming - for example if
a filename is already in use, it will have a number (or a random id) appended
to the end. Using metadata, it could be possible to store the original filename:
this tag is probably desirable anyway because the file extensions have
already been chopped off. If output from exiftool is processed via
process-metadata.php, this information would be stored already but we cannot
make that assumption. A basic set of metadata should be added when an image is
added to the database using process-filelist.php or another means. If an
automatic id is used, there must be a way to determine which id to use next,
preferably without requiring another database table but still able to
differentiate between numbers in the filename and numbers added because of
a filename clash.

Useful tags:
- original file name
- import date and time --> how do we want to store this? UNIX timestamps are
  simple, but the meta value column is a text column and the timestamps aren't
  very user friendly

Summary:
[high] Write metadata containing original filename, import date/time when a
       file is imported
[low]  If filename collision occurs, append a counter or random id to filename