At 5:56p -0400 on Sun, 13 Jul 2008, Sol wrote: > I do not see any consistent behavior. It could mean one of the following: > > order by: > > <Genre> than <Artist> (in reverse order) than <Album> > <Genre> than <Album> than <Artist> > <Genre> than <Album> than <Artist> (in reverse order)
I think *something* should happen for the user if the user clicks. I just don't know what ... [... head scratching ...] Let me return to CS1: the behavior for which I'm hankering is basically a stable sort. That is, if, on any given key there is a duplicate entry, then, when sorting them, preserve the original order in which they were found. For our example, this then means that when the user clicks <Genre>, <Artist>, <Album> <Album> is the final click, so the entire viewable set is organized by <Album>, <Artist>, <Genre>. Going from this, what would happen in the 4-click example you suggested: <Genre>, <Artist>, <Album>, <Artist> Now, the *first* <Artist> is effectively superfluous, but since the result of a click is a sort, the user interface is the same: click once on a column, sort it ascending. Click twice in a row and sort it descending. Clicking three times is the same as sorting once. When I sort by the next column, I don't sort with the previous column specifically in mind. However, the sort is stable so the sub-ordering behavior that I initially suggested is there. I believe this would effectively side-step the user-communication issue that you rightly brought-up since the list organization happens per-click, and you're only showing the most significant ordering. I still suggest that Rhythmbox use SQLite, though, so how to reconcile this with SQL? I don't know yet. Kevin _______________________________________________ rhythmbox-devel mailing list [email protected] http://mail.gnome.org/mailman/listinfo/rhythmbox-devel
