pull request

[Linus Torvalds]

On Fri, Aug 17, 2012 at 2:35 PM, Dirk Hohndel <dirk at hohndel.org> wrote:
>
> this fixes the selection oddities that have been introduced with the
> dive trip based grouping.

I found a new selection oddity.

Try this:

 - collapse all dives
 - select one dive-group
 - shift-select another dive group two entries down (so you have three
dive groups selected)

   SUCCESS. Everything works the way you expect, and you can see that
the stat count matches the sum of the three dive groups. Expand it
all, and you see all the dives you expect to be selected really are
selected. Goodie. This is great for doing things like "select all
dives of 2011" to get statistics.

Now, try this:

 - expand all dives
 - select one dive-group (everything fine so far, you get exactly the
expected behavior)
 - shift-select another dive group two entries down

You'd *expect* that what happens is exactly the same thing as before,
and it only adds to the selection.

Except nope, that's not what happens at all.

What actually happens is that the first dive-group gets *unselected*,
except for the very first dive in that group (== last dive in the list
visually). So instead of the dives of all three dive-groups being
selected, you have two dive groups selected and just one dive from the
first group.

Very odd.

I'll try to take a look and read the source and figure it out, but I
suspect that since you know the selection code, you'll just go "Argh,
I know exactly what's going on, I forgot another gtk corner case".

               Linus