Merge lp://qastaging/~3v1n0/unity/lim-panel into lp://qastaging/unity

8 +BuilderWrapper& BuilderWrapper::add(char const* name, long int value)
9 +{
10 + g_variant_builder_add(builder_, "{sv}", name, g_variant_new_int64(value));

long is 32-bit on i386, if you really need 64bit value this should be changed to "long long".

> 8 +BuilderWrapper& BuilderWrapper::add(char const* name, long int value)
> 9 +{
> 10 + g_variant_builder_add(builder_, "{sv}", name,
> g_variant_new_int64(value));
>
> long is 32-bit on i386, if you really need 64bit value this should be changed
> to "long long".

Right. I've added wrappers for long long too, anyway I'm keeping the variant as a gint64 also for long int, to make sure it's working also on 64bit...

"The diff has been truncated for viewing. "

Try to avoid doing too much rewriting in any one branch :)

466 PanelIndicatorEntryView::~PanelIndicatorEntryView()
467 {
468 - on_indicator_activate_changed_connection_.disconnect();
469 - on_indicator_updated_connection_.disconnect();
470 - on_panelstyle_changed_connection_.disconnect();
471 - g_signal_handler_disconnect(gtk_settings_get_default(), on_font_changed_connection_);
472 - if (texture_layer_)
473 - delete texture_layer_;
474 + // Nothing to do...
475 }

The empty destructor's definition can be removed (unless its virtual)

+ if (!wm->IsExpoActive() && !wm->IsScaleActive())
490 + {

This probably doesn't make sense living where it is. The Panel views should block all clicks if if the screen is grabbed by compiz period (eg, add a wrapper method in WindowManager around CompScreen::otherGrabExist (NULL);)

3181 + /* FIXME, this can be removed when window_unmapped WindowManager signal
3182 + * will emit the proper xid */
3183 + Window xid = bamf_window_get_xid(reinterpret_cast<BamfWindow*>(view));
3184 + OnWindowUnmapped(xid);

Can you explain this? It could be emitting the frame xid, and that needs to be fixed there :)

nvm found it

3391 + // FIXME: compiz doesn't give us a valid xid (is always 0 on unmap)
3392 + // we need to do this again on BamfView closed signal.

Great. I'll fix that in compiz then. (This is probably the case when the window is destroyed and the xid is changed to zero, and then the unmap signal happens)

The window management stuff looks great.

> "The diff has been truncated for viewing. "
>
> Try to avoid doing too much rewriting in any one branch :)

Yes, I know, but things often come in mind when doing it, and as I'm touching a lot of panel stuff, I'd prefer to keep all here.
You can just use bzr send lp:unity -o branch-changes.diff to check them locally ;)

> The empty destructor's definition can be removed (unless its virtual)
>
> + if (!wm->IsExpoActive() && !wm->IsScaleActive())
> 490 + {
>
> This probably doesn't make sense living where it is. The Panel views should
> block all clicks if if the screen is grabbed by compiz period (eg, add a
> wrapper method in WindowManager around CompScreen::otherGrabExist (NULL);)

Well, if the dash grab isn't considered by that, I could try to do it.

> 3391 + // FIXME: compiz doesn't give us a valid xid (is always 0 on unmap)
> 3392 + // we need to do this again on BamfView closed signal.
>
> Great. I'll fix that in compiz then. (This is probably the case when the
> window is destroyed and the xid is changed to zero, and then the unmap signal
> happens)

Let me know when your fix will be landed, so we can remove this.

> The window management stuff looks great.

Cool! ;)