Re: [Pharo-dev] OS-Windows and UnifiedFFI

[Esteban Lorenzano]

On 24 Jun 2016, at 09:53, Nicolai Hess <nicolaih...@gmail.com> wrote: I’m sorry… I do not understand. This is solved in UFFI since some months, and Torsten’s OSWindow uses it.  Sorry then it is my fault. I know that Torsten already ported OSWindow to UFFI, and that most things are already working. UFFI provides FFIConstantHandle type to deal with windows HANDLE types, as comment says:  I represent a constant HANDLE, as described in  *Windows MSDN>https://msdn.microsoft.com/en-us/library/windows/desktop/ms724457(v=vs.85).aspx* A ==HANDLE== is a special kind of external object who is accessed through numbers, therefore and ==ExternalAddress== is not appropriate to describe it (since they are constants and external addresses represents disposable spaces from memory). Is not clear this is necessary outside Windows, but according to documentation they are somekind analogous to unix's File Descriptors (but with some remarkable diferences, as documented *here>http://lackingrhoticity.blogspot.fr/2015/05/passing-fds-handles-between-processes.html*. Example:  [[[ HWND := #FFIConstantHandle. self ffiCall: #(HWND GetActiveWindow()) ]]] is there anything I’m missing? I don't know what else should I do, I already posted the example. The code I provided just don't work (DNU on isExternalAddress, see other mail), and I have no idea how to make it work. Yes, maybe I don't know enough about ffi for this type of the function argument passing. the example is a simplification, you cannot put both definition and call in same method… find attached a working example of what you want.

HandleExample.st
Description: Binary data

(actually I tried to port some NBOpenGL code to UFFI, for this I need to register a Window Class with WinWndClassEx (to create a gl context). So if this *is* already working, I need to look what I have done wrong). thanks for your response. nicolai   Esteban