Newsgroups: comp.lang.scheme
Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!think.com!snorkelwacker.mit.edu!bloom-beacon!dont-send-mail-to-path-lines
From: gyro@cymbal.reasoning.COM (Scott Layson Burson)
Subject: Ports & GC
Message-ID: <9104291751.AA10911@cymbal.reasoning.com.>
Sender: daemon@athena.mit.edu (Mr Background)
Reply-To: Gyro@reasoning.com
Organization: The Internet
References: <480@data.UUCP>
Date: 29 Apr 91 17:51:47 GMT
Lines: 29

   Date: 26 Apr 91 15:40:30 GMT
   From: Ken Dickey <data!kend@bloom-beacon.mit.edu>

   [Answering a question about whether garbage ports are closed automatically]

    Implement finalization objects.  I did this in a local copy of the
   Gambit runtime (v1.5) by implementing weak-pairs (the car goes to nil
   when collected).  The finalization procedure registers each object to
   be finalized with a thunk which does the cleanup.  The object is in
   the car of a weak pair, the thunk in the cdr.  After a collection, the
   weak-pairs are scanned and any who's car is nil is thrown away after
   the finalization thunk is invoked.  {The rules are that such a thunk
   does not cons and does not invoke random continuations.}

   A finalization service is also helpful for freeing storage for dumb
   allocators like C's malloc.  Weaks/populations, etc. are helpful in a
   variety of ways.

Somebody on this list recently alluded to "the well-known difficulty of
defining UNWIND-PROTECT in the presence of firstclass continuations".

I have long thought that what UNWIND-PROTECT ought to mean is that the
cleanup actions are performed when the continuation in which they were
established becomes garbage.

As Ken points out, this really isn't all that hard to implement.

-- Scott
Gyro@Reasoning.COM