Now that my rewrite of pidl has been finished, I am wondering what the future for pidl will be.
Looking at the new structure of the source code, I notice that pidl's structure is now very simple. First, it generates a datastructure representing the IDL file; then it generates new datastructures by walking the first one recursively. In the last step, it recursively walks a datastructure generating output.
Things like this are very easy to do in Haskell, using a trivial parser and a couple of fold functions, as I learned at the course on grammars and parsing at uni.
The parsing step is probably the most difficult one, i.e. going from a file to something like this:
type Property = (String, [String]) type EnumMember = (String, String) data Type = Struct (Maybe String) [DataMember] | Union (Maybe String) [Maybe DataMember] | Enum (Maybe String) [EnumMember] | Reference String | Interface String (Maybe String) [Property] [InterfaceMember] type Typedef = (String, Type) data InterfaceMember = Function String [Property] Type [DataMember] | Typedef String [Property] Type | Const Type String String | Type Type type DataMember = (Type, Int, String, [Property], [String])
Ah well, back to more important things now...Go Top