Hello,
Procedures are stateless entities that are copied without distributed
computation because they are constants. So the procedure is executed on
the client process and success is displayed at standard output. But
messagePort is unreachable because network transparency is not applied to
it.
As the procedure is executed on client process, messagePort should be
accessed by a pickle in it.
All of this should be validated by a mozart's guru because I'm not sure of
it ...
Thibault Germentier
> hello,
>
> while implementing distributed apps, i observed a behaviour that i can't
explain.
> here's a simplified code that preserves the functionality needed to
reproduce the effect.
>
> here is a simple server that collects messages from clients using a
stream and a port:
>
> functor
> import Connection Pickle System
> define
> Messages
> MessagePort = {Port.new Messages}
> proc {Post post(In Out)}
> {Send MessagePort In}
> Out = success end in
> {Pickle.save {Connection.offerUnlimited Post} post}
> for Message in Messages do {System.show Message} end
> end
>
> here is a simple client that sends a message to the server:
>
> functor
> import Connection Pickle Application System
> define
> Post = {Connection.take {Pickle.load post}}
> Response = {Post post(hello $)} in
> {Wait Response}
> {System.show Response}
> {Application.exit 0}
> end
>
> the code roughly corresponds to the examples from ch. 11 in ctmcp,
(e.g., sec. 11.3).
>
> the server and the client are compiled. the server is started, and then
the client attempts to send a message, and exits afterwards; the server
does not display anything, and the process starts spinning (takes up all
cpu time) as soon as the client is dead.
>
> if i add a delay to either the client code or the server code so that
the client has to wait a moment (say,10sec) before it terminates, the
message is displayed by the server, but then the server starts spinning
again after the client terminates. when the server is spinning,
starting a new client process blocks.
>
> if i execute the client code (without the functor part) in an opi, all
works fine until i terminate the opi's oz process. then the server
starts spinning again. i can connect and successfully communicate with
the server again, even though it is spinning.
>
> however, if the server is spinning after a command line client has
terminated, then a new opi client blocks as well. (but if there was an
opi client already connected when a cl client blocks, it is still able
to communicate with the server.)
>
> what's the problem?
>
> vQ
>
> --
> -------------------------------------------------------------------------------
Wacek Kusnierczyk, MD PhD
>
> Email: [EMAIL PROTECTED]
> Phone: +47 73591875, +47 72574609
>
> Department of Computer and Information Science (IDI)
> Faculty of Information Technology, Mathematics and Electrical
Engineering
> (IME)
> Norwegian University of Science and Technology (NTNU)
> Sem Saelands vei 7, 7491 Trondheim, Norway
> Room itv303
>
> Bioinformatics & Gene Regulation Group
> Department of Cancer Research and Molecular Medicine (IKM)
> Faculty of Medicine (DMF)
> Norwegian University of Science and Technology (NTNU)
> Laboratory Center, Erling Skjalgsons gt. 1, 7030 Trondheim, Norway Room
231.05.060
>
> -------------------------------------------------------------------------------
>
> _________________________________________________________________________________
mozart-users mailing list
> [email protected]
> http://www.mozart-oz.org/mailman/listinfo/mozart-users
>
_________________________________________________________________________________
mozart-users mailing list
[email protected]
http://www.mozart-oz.org/mailman/listinfo/mozart-users
