On Sep 16, 2013, at 1:21 PM, Ken Thomases wrote: > On Sep 16, 2013, at 12:55 PM, Alexandre Julliard wrote: > >> It's better to avoid adding entry points that don't correspond to >> Windows APIs. Instead you should request the info when you need it. > > Cocoa makes use of the size limits "spontaneously", not just at the time when > the user begins to resize a window. On OS X 10.7 and later, the user can > resize windows by their edges and the cursor changes to show the available > operations. So the limits need to be set in advance and updated whenever the > app would respond differently to WM_GETMINMAXINFO. Since I have no way of > knowing in advance when it might do so, I figured I would either have to poll > (more or less) or piggyback on user32's queries. > > Also, it's not clear to me that apps will tolerate WM_GETMINMAXINFO occurring > at times when Windows would not send it.
Well, the first "app" that doesn't tolerate it is the user32 test suite. Other than when a user starts to resize a window, any other time I could think of to send WM_GETMINMAXINFO causes test failures. For now, I've sent new versions of the patches which only get the info at the start of a resize. Until then, Cocoa won't know the size limits and may display incorrect resize cursors at the window edges. Any suggestions on other times/ways I could query the min/max info without generating messages that will break the tests? -Ken
