[PATCH] Remove unused variable from File class>>#checkError:

Hi,

while looking into why DatagramSocket loocal: '8.8.8.8' port: '23000' is
not giving an exception I looked into the file class, the errors
variable appears to be unused and I think it is safe to remove it.


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

On 06/16/2010 02:31 PM, Holger Hans Peter Freyther wrote:
> Hi,
>
> while looking into why DatagramSocket loocal: '8.8.8.8' port: '23000'

st> Sockets.DatagramSocket local: '10.34.30.119' port: 23000
Sockets.DatagramSocket[10.34.30.119:23000]
st> Sockets.DatagramSocket local: '10.34.30.119' port: '23000'
Object: IPAddress new "<0x7f5a05e803c0>" error: port out of range

?

Paolo

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

On 06/16/2010 08:37 PM, Paolo Bonzini wrote:
Sorry, of course we need to pass the port as a number... the point is
more that binding to googles public ip on my system should fail and it
didn't.

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

On 06/18/2010 03:23 AM, Holger Hans Peter Freyther wrote:
Why?  You just won't get any datagram, AFAIU.

Paolo

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

On 06/18/2010 05:34 PM, Paolo Bonzini wrote:

>> Sorry, of course we need to pass the port as a number... the point is
>> more that binding to googles public ip on my system should fail and it
>> didn't.
>
> Why?  You just won't get any datagram, AFAIU.

Sure, you will not get any datagram and the read should fail pretty fast.

st> Sockets.DatagramSocket local: '8.8.8.8' port: 23000
Sockets.DatagramSocket[0.0.0.0:0]

generates the following syscalls:

bind(3, {sa_family=AF_INET, sin_port=htons(23000),
sin_addr=inet_addr("8.8.8.8")}, 16) = -1 EADDRNOTAVAIL (Cannot assign
requested address)
rt_sigprocmask(SIG_BLOCK, ~[QUIT ILL ABRT BUS SEGV RTMIN RT_1], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
getsockopt(3, SOL_SOCKET, SO_ERROR, [0], [4]) = 0
rt_sigprocmask(SIG_BLOCK, ~[QUIT ILL ABRT BUS SEGV RTMIN RT_1], [], 8) = 0
rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
getsockname(3, {sa_family=AF_INET, sin_port=htons(0),
sin_addr=inet_addr("0.0.0.0")}, [16]) = 0

and I think the failure of the bind should be propagated as an
exception, it does not as the SOL_ERROR is not reporting any error (okay
I didn't verify that).





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

On Sun, Jun 20, 2010 at 17:22, Holger Hans Peter Freyther
<[hidden email]> wrote:
Yes, that's correct.  Maybe it's a kernel bug.

Paolo

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