[KPhotoAlbum] Very experimental image-grouping patch

[Paul Fleischer]

2007/11/29, Tero Tilus <tero at tilus.net>:
> Very much true.  I do not really have any even medicore ideas on how
> to communicate this relationship to ... oh, wait ... I have!  How
> about (the already presented idea) that the pics having a relationship
> would have some overlaid icon to indicate it.  Clicking the icon would
> go to "item neighbourhood" view, where you could see the selected
> image and all related 1 (to configurable n steps) further.  Relations
> could be presented with arrows and maybe some day give up the grid and
> use Kamada-Kawai (like Graphviz NEATO) or like to arrange photos.
>
> Well.  Thinking about it more closely, it is open to debate if that
> was even medicore.  :)

I think it's a good idea. Right now I'm working on improving the way
children are show, using the current grid-view. And actually, I will
try to implement many-to-many as well :-)

> > But in practice I think grouping images is much easier to deal with,
> > implementation and interaction wise.
>
> Well, you already did the master-derivate relationship.  ;)
> User interaction is an issue, though.

:-)

> At least it's not counterintuitive to me. ;) That's exactly why I
> wanted to emphasize that hiding images and handling master-derivate
> relationship are separate issues.  You can not judge what to show only
> by the master-derivate relation.  You might have master and a
> retouched derivate and you might very well want to show one, the
> other, both or neither of them.  You might very well want to hide
> items (I'll work with this some day and don't wanna show it before
> that) even if they do not have any relationships.  That said, it's of
> course desirable to have functionality to handle commonly repeatin
> hiding patterns with respect to relationships.

Agreed. I might hack in some item-hiding just for proof-of-concept...
unless I'm stepping on someones toes by doing that.

I guess the conclusion of all this is that I'm working towards the
many-to-many relationship concept. Do please remember that all this is
just for proof-of-concept. We can always change our minds at a later
point.

/Paul