Merge lp://staging/~rafalcieslak256/ubuntu-accomplishments-viewer/uav-display-modes into lp://staging/ubuntu-accomplishments-viewer

I don't claim to understand all the GTK stuff here, but the theory makes sense to me. I do have a few questions.

What about the XXX on line 178? I don't follow what the optimization is?

Q: Why is the code in on_window_resized() all moved into a string?

I see this note about making these global, I think they'd make sense as global:
951 + today = datetime.date.today()
952 + margin_today = datetime.timedelta(days = 1)
953 + margin_week = datetime.timedelta(days = 7)
954 + margin_month = datetime.timedelta(days = 31)
955 + margin_sixmonths = datetime.timedelta(days = 180)

Why did you remove all these lines? What did they do?
 <property name="use_action_appearance">False</property>

Great thanks for taking time to look as this, Matt! :)

To answer your questions:

The XXX note in this diff's line 178 is an annotation for following prepare_models() call. The prepare_models() function is the only time-consuming bit now, as it initializes both treemodels. The call in line 178 is within function that runs when a new trophy is received. The point is that it makes no sense to clear both treemodels and recreating them from scratch when a new trophy is awarded, instead, we might just remove that new trophy from opprotunities list and add it to mytrophies. We do know the accomID of that new trophy, as daemon provides us with that data, so it's just a matter of implementing support for such action. It would be just faster.

Code in on_window_resized() is moved to a string because I wanted to remove it, but moving it to a comment looked to me as a more appropriate way of disabling it, so that the original code is left in case we wanted to reuse it some time. Currently it's not needed, as GTK now correctly manages iconview's width, and we don't need any silly workarounds to have the window properly resizing. Moreover, that trick we used as a workaround based on clearing and recreating all treemodels - and that's what this branch wants to avoid, since it's a very time-consuming operation.

I have not changed any of the "use_action_appearance", I am pretty sure Glade has edited that on it's own. Most likely a new version of glade finally recognizes that there is not much sense in specifying explicitly that "use_action_appearance" should be false, as that's GTK's default behavior. Thus, the removal of these lines shouldn't change anything.

If you want to, I can provide some more detail on how the GTK stuff works here.

I think you should just remove the on_window_resized code, we can get it back via bzr history if we need to, otherwise it looks good. I don't figure you're going to get an in depth review from Jono anytime soon, maybe Michael Hall could review it? Personally I think you should just push it in, we'll need to do extensive testing on this code before 0.4 anyway and now is the time for major changes like this.

Thank you for the review. I'll push these changes, it's the first major thing for 0.4 so we'll have lots of time to test it afterwards.