On Tue, Aug 07, 2007 at 04:57:30PM -0700, Michael Vanier wrote:
Other comments:
-- I find the way that Screen wraps a workspace somewhat icky, though I can't think of a better alternative. It feels like it compromises the symmetry of the StackSet data structure. Maybe I'm just being too picky here.
Agreed, it's very ugly. More symmetric options don't seem to be very popular, thought. It seems to me that the "right" way to do things is to use hierarchical Stacks of one sort or another, and handle the arranging of workspaces on screens in a manner symmetrical to the handling of windows on workspaces (and in a Layout like Combo, or when there are float windows forming a sub-workspace, the handling of sub-workspaces within workspaces). It seems best to me to have a single data structure which is used to describe "a set of things, one of which has focus." Currently, we've got two: StackSet describes a set of Screens, one of which has focus, and Stack describes a set of Windows, one of which has focus. StackSet also describes a set of Workspaces, one of which has focus implicitly, via the mapping between Screens and Workspaces.
-- Why do we need the StackOrNot type alias? What's wrong with Maybe (Stack a)? IMO the latter is more readable, or more readable when you use Maybe monad functions.
It's my fault, and in retrospect it wasn't such a hot idea. My thought was that StackOrNot would indicate that Nothing doesn't indicate an error, just that there are no windows present. I wouldn't object to a patch removing StackOrNot. -- David Roundy Department of Physics Oregon State University