[KPhotoAlbum] Associating multiple tags with one area?

[Johannes Zarl]

I guess that GUI- and workflow-wise the feature can be implemented in a 
relatively straight-forward way:

The context-menu on an area would only need minor rephrasing:
1. Add [last added tag] to area
2. Add tag to area -> [Submenu with not-associated tags]
3. Remove tag from area -> [Submenu with associated tags]
4. Remove area from image
(Ignoring the face detection menu entries)

The replace operation should be a seldomly used one, and isn't currently 
implemented, either. Thus, the workflow "add new tag, then remove old tag" 
should work sufficiently enough.

Database-wise, I don't really know how much this will affect efficient 
handling of areas. As Hugues suggested, the easiest way to implement multiple 
tags per area would be to store duplicate areas in the database, and deal with 
them in the UI code.
This way we would probably miss out on some opportunities for optimisation, 
but otherwise I'm all in favour of not changing the database format.

> I
> noticed that there's some sort of extension/plugin mechanism. Would that be
> capable of changing the UI to a great enough degree to implement this?

You probably mean the KIPI-Plugins. You can't really change the UI through 
these plugins. The idea behind KIPI is to share common functionality between 
KDE image-related programs.

> Thinking about this, maybe there could be a button to 'stack' areas
> together? 'Stacked' areas could have some link or identifier that could be
> generated simply by detecting perfect overlap, which should be extremely
> rare in existing circumstances. With some kind of identifier or link,
> making edits affect all areas in a stack shouldn't be too difficult. Once
> in a stack(Which could be known by checking for this theoretical 'link'),
> the menu could change to add new tags to the set, or remove existing ones
> from it.

I'm not sure about stacking areas. It seems to add more complexity for little 
gain.

> I've got a copy of the git repo now, so I'll be taking a look at the
> relevant code sometime later. Hopefully I'll be able to at least try some
> of these, and see for sure if they would actually work.

Sure. If you need comments or help, just ask...

Cheers,
  Johannes