Help needed for gst-remote

Hi all,

I have compiled GNU Smalltalk version 3.1-bab472a and I have installed it into
$HOME/install/gst. Now I'm trying to follow the blog post about seaside and
GNU smalltalk[1].

In one terminal I type:
  gst-remote --daemon

The effect is that it goes to the background, prints a message that it is
started and I see that someone bound to 0.0.0.0:12345

In another terminal I attempt to type:
  gst-remote --eval "3+3"

The effect is that it sits there and will die with a connection timeout after a
while.I'm running this on a 2.6.33-rc8 with a current Ubuntu version of the
Ubuntu Lucid development series.

So far I have tried to patch the Remote.st to print a message after there has
been a connection, when using telnet/nc on the above port I see this message
printed, when using gst-remote --eval 3+3" I don't.

has anyone else seen this? what am I doing wrong? Could the number of
interfaces play a role? Should I try to see which address the client tries to
connect to?

regards
        holger




[1] http://smalltalk.gnu.org/blog/bonzinip/seaside-development-gnu-smalltalk


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk

On 03/03/2010 01:11 PM, Holger Hans Peter Freyther wrote:
> So far I have tried to patch the Remote.st to print a message after there has
> been a connection, when using telnet/nc on the above port I see this message
> printed, when using gst-remote --eval 3+3" I don't.
>
> has anyone else seen this? what am I doing wrong? Could the number of
> interfaces play a role? Should I try to see which address the client tries to
> connect to?

Yes, try "strace gst-remote --eval 3+3".

Thanks!

Paolo


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk

On Wednesday 03 March 2010 15:07:54 Paolo Bonzini wrote:

> >
> > has anyone else seen this? what am I doing wrong? Could the number of
> > interfaces play a role? Should I try to see which address the client
> > tries to connect to?
>
> Yes, try "strace gst-remote --eval 3+3".


connect(4, {sa_family=AF_INET, sin_port=htons(12345),
sin_addr=inet_addr("10.88.73.112")}, 16) = -1 EINPROGRESS (Operation now in
progress)


the weird part is that this ip is not any address on my interface, not even of
my current networks. I will poke Remote.st after dinner...

thanks


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk

On Wednesday 03 March 2010 19:52:58 Holger Hans Peter Freyther wrote:

>
> the weird part is that this ip is not any address on my interface, not even
> of my current networks. I will poke Remote.st after dinner...

I don't pretend to understand everything. The weird IP Address I was seeing is
the "TCP.IPAddress anyLocalAddress" address. So apparently I have a bogus
entry in /etc/hosts which was picked up.

My two questions. Wouldn't it make sense for the Remote.st to use the
loopbackHost selector instead of the anyLocalAddress? And the second one is...
instead of getting the hostname and then trying to resolve it, couldn't we
iterate over the interfaces and pick the ips from them?

regards
        holger



_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk

On 03/03/2010 07:52 PM, Holger Hans Peter Freyther wrote:
> connect(4, {sa_family=AF_INET, sin_port=htons(12345),
> sin_addr=inet_addr("10.88.73.112")}, 16) = -1 EINPROGRESS (Operation now in
> progress)

That's expected, it is not an error.  gst will be signaled later when
the connect is complete and it suspends the Smalltalk process in the
meanwhile.

I'm sure the strace was longer. :-)

Paolo


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk

On Thursday 04 March 2010 10:04:45 Paolo Bonzini wrote:

> I'm sure the strace was longer. :-)

You are right. I normally hate when users decide to filter the information and
decide to know better. The connect had all the information I considered
relevant. This is a slight duplication of the last mail. The chosen IP address
was just plain wrong... and this explained the Remote.st timing out.

thanks
        z.


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk

> My two questions. Wouldn't it make sense for the Remote.st to use the
> loopbackHost selector instead of the anyLocalAddress?

Couldn't the loopback interface be down?

> And the second one is...
> instead of getting the hostname and then trying to resolve it, couldn't we
> iterate over the interfaces and pick the ips from them?

Yes, possibly.

Paolo


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk

On 03/04/2010 01:19 PM, Holger Hans Peter Freyther wrote:
> I normally hate when users decide to filter the information and
> decide to know better. The connect had all the information I considered
> relevant.

... but what came before might have given a clue as to why the address
was chosen.

Glad you figured it out!

Paolo


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk