On Apr 4, 2008, at 1:53 AM, Michael Sperber wrote:

> Now, I realize this isn't explicitly covered in the spec, but  
> shouldn't
> an implementation flush open ports upon exit?  This is pretty  
> widespread
> practice both in Scheme and other languages, and at least the examples
> seem to imply that this was the intention.

If it's not specified explicitly, I would leave it to R7RS.

I ran into some corner cases when I tried to do this.  What happens  
for example if you try to flush a port but an IO error occurs?  Can  
an exception be raised between the time you exit (explicitly or  
implicitly) and the time the process actually exists?  One can argue  
either way I think.

As the spec stands, it's more portable and predictable for the users  
to flush their ports themselves (and that's compatible with auto  
flush on exit).  I don't think you should have an example in the spec  
that relies on this unspecified behavior (even if all implementations  
did so anyways).

Aziz,,,

_______________________________________________
r6rs-discuss mailing list
[email protected]
http://lists.r6rs.org/cgi-bin/mailman/listinfo/r6rs-discuss

Reply via email to