[gambit-list] output ports created with ##open-predefined not being garbage collected

Jörg F. Wittenberger Joerg.Wittenberger at softeyes.net
Sat Apr 11 14:16:55 EDT 2020


On Sat, 11 Apr 2020 13:08:23 -0400
Marc Feeley <feeley at iro.umontreal.ca> wrote:

> > On Apr 11, 2020, at 1:01 PM, Jörg F. Wittenberger
> > <Joerg.Wittenberger at softeyes.net> wrote:
>...
> Are you communicating with UDP?  If that is the case then maybe you
> can use Gambit’s UDP ports directly:

No, it's abstract unix domain sockets in the case at hand.

But I'm more interested to learn how to get the low level things
straight in principle.

> (open-udp port-number-or-address-or-settings)

At another currently questionable code spot (after all, may code fails
currenlty with confused malloc in short order:-/ ) I actually tried to
used gambit's UDP.

But I failed to find out how to bind the outgoing port to the one I'm
listening on for the reply.  So I went back to use gamsock again.

/Jörg

> This procedure opens a socket for doing network communication with
> the UDP protocol. The default value of the direction: setting is
> input-output, i.e. the Scheme program can send information and
> receive information on the socket. The sending direction can be
> closed using the close-output-port procedure and the receiving
> direction can be closed using the close-input-port procedure. The
> close-port procedure closes both directions.
> 
> The resulting port designates a UDP socket. Each call to read and
> udp-read-subu8vector causes the reception of a single datagram on the
> designated UDP socket, and each call to write and
> udp-write-subu8vector sends a single datagram. UDP ports are a direct
> subtype of object-ports (i.e. they are not character-ports) and read
> and write transfer u8vectors. If read is called and a timeout occurs
> before a datagram is transferred and the timeout thunk returns #f
> (see the procedure input-port-timeout-set!) then the end-of-file
> object is returned.

Just re-read this.  Still no clue.





More information about the Gambit-list mailing list