On Fri Oct 13 08:34:39 2023 +0000, Alexandros Frantzis wrote:
I think that if we want semantic and naming consistency the best field to use would be `data->whole_rect` (populated with `visible_rect`). As you noted in an earlier comment, what we need for our current operations is in fact the rect of the window that will be displayed by the native system, which is what `data->whole_rect` is meant to hold (IIUC). At this time, `whole_rect`/`visible_rect` happens to always match `window_rect`, so we could also accept this assumption and just use `data->window_rect` (populated by `window_rect`). Is this what you are proposing? This is what we had before with the use of `NtUserGetWindowRect`, but I was under the impression that, in addition to getting rid of the `NtUser` call, we also wanted to remove the aforementioned implicit assumption (but perhaps I misunderstood). I would have a slight preference for introducing the fields on demand, but I am fine if you see some benefit in adding them all now.
Because there's no difference at this point, I think using `window_rect` would be better. And because the driver doesn't implement `visible_rect`, you wouldn't even have to fill it.
I would have a slight preference for introducing the fields on demand,
Fine by me.