[clean-list] output of startIO

[Peter Achten]

Dear Erik,

This is right. The point is that in the 0.8 I/O library you basically 
had just a single interactive process (with nested interactive 
processes). If you have a single process, then it is no problem to 
define what its final state is. Object I/O, on the other hand, supports 
an arbitrary number of arbitrarily dependent interactive processes. If 
each interactive process creation function would return the final state 
of the created process, then you can define an interactive process that 
decides to create a child process, and then tries to evaluate the result 
state value of its child. Consequently, it blocks until its child 
terminates. This is not what an interactive process should do: each 
state transition should terminate timely. In addition, Object I/O 
supports message passing between interactive processes, hence it is 
always possible to set up communication between a parent and its 
children processes. At the root this is indeed not possible, hence you 
need to resort to something like writing the final state to file or 
something, using either overloaded printing, generic printing, or dynamics.

With kind regards,
Peter Achten

erik.zuurbier at tiscali.nl wrote:
> In the 0.8 I/O-library we had the StartIO function whose output-type 
> was a tuple. In one of the elements of the tuple, one could return the 
> final state of the interaction.
>
> In the Object-IO-lib the output-type of startIO is just *World. It 
> does not seem to be possible to pass anything to the surrounding 
> program when the interaction finishes. Is there a way to do that 
> anyway?
>
> Erik Zuurbier
>
>
>
> _______________________________________________
> clean-list mailing list
> clean-list at science.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/clean-list
>