[PATCH] When only a dive trip is selected, select all the dives in the trip

[Linus Torvalds]

On Mon, Feb 18, 2013 at 3:43 PM, Linus Torvalds
<torvalds at linux-foundation.org> wrote:
>
> This does a final pass after all the selection logic, and notices if we
> have dive trips that are selected, but that have no dives in them
> selected.  In that case, we assume that the user wanted to select all
> dives in that trip.

Btw, to see just how much of a "magic special case" this is, do this:

 - select a dive trip (all dives in that trip get selected)

 - one by one, de-select each individual dive (ctrl-mouse-click)

 - as you de-select the last individual dive, we now have the
situation where the dive trip is selected, but no dives under the trip
- so our special-case logic triggers, and all dives get re-selected.

And yes, I can fix this, by having the logic be that we *only* do the
whole special case if the dive trip itself was in the set of entries
that got a "selection might change" callback. But from a UX
standpoint, I suspect we don't even need to do this all, since the
user really has to *work* at triggering the special case this way.

My point really is that it's a magic special case, and while it fixes
the normal behavior, the magic rules really are magic rules, and you
can see them in action by doing odd things to trigger them other ways.

So I'm not arguing against the patch (I think it does what people want
in all the normal cases), I'm just pointing out how odd things are
when we have to fight the native gtk selection rules, and make up
these kinds of extra rules on top.

                     Linus