A few random I/O proposals

[philbin@research.nj.nec.com (James Philbin)]

   Date: Thu, 28 May 92 09:06:54 -0400
   From: jmiller@crl.dec.com
   X-Mts: smtp

   > Since Scheme has no error/condition system yet, and since having
   > OPEN return #f for failure is an imperfect solution for reasons
   that > have been well covered, how about having OPEN accept a
   failure > continuation argument?

   Bravo, I think.  I'd alter that just slightly.  I'd like a
   signature of something like:

        (open-input-file file-name) as is.

        (open-input-file file-name error-handler) if the open fails
        (i.e. if the one argument version would signal an error),
        error-handler is called with an implementation-specific
        argument describing the reason and the result of this call is
        returned from open-input-file (i.e.  open-input-file "tail
        recurses" into error-handler).

   With this change from your suggestion, I think Dick might be happy.
   And, for the record, Dick wasn't the only one who wanted CALL/CC
   removed.  I voted that way as well.

   --Jim

Exception handlers still want the ability to "throw" to some
previously saved continuation.  So you still need CALL/CC or at least
"downward only" continuations, i.e. CATCH.

    Jim