I'm back, it was a short trip but I had a lot of fun talking to people, learning what they are doing with Windows, and yes, learning more about Borland and Delphi. I also learned a bit more about Borland's current thinking around WPF -- none of this is set in stone, and in any case I obviously I don't work for Borland so I can't make any commitments about what they are going to do, but I can repeat what Danny Thorpe said at his talk:
That last part was definitely something Danny thought a possibility. It's an interesting and workable approach. We considered doing it for Windows Forms, but ultimately decided to use an hwnd interop-based implementation instead. The good part about the "reimplement on top of WPF" approach is that there's no airspace restrictions. But for Windows Forms it wouldn't have been compatible enough for our needs -- a lot of Winforms third party components depend on the hwnd-ness, and such a reimplementation would change a lot of subtle semantics and event orderings. But WinForms is not VCL, and Borland has a lot of experience with this sort of change (eg. VCL for win32 -> VisualCLX, aka VCL implemented on Qt), so I'm looking forward to it and a chance to write more about VCL-WPF interop!
(And for everyone who saw my talk at Borcon, I apologize for showing code in notepad but yes, I really did write that code -- I have Delphi installed on my desktop machine, just not on my laptop)