Discussion:
Problem with running examples
Aseem Rastogi
2013-10-14 17:55:02 UTC
Permalink
Hi All:

I am trying to install VIFF and run tests on following configuration:

Mac OS 10.8
python: mac port python27 @2.7.3_1
gmpy: mac port py27-gmpy @1.17_0
twisted: mac port py27-twisted @12.0.0_0

When running millionaire's problem, I am getting this error:

Seeding random generator with random seed 1945
Using SSL
Listening on port 4002
Will connect to <Player 3: localhost:4003>
I am Millionaire 2 and I am worth 83 millions.
*** bad data from Player 3: unpack requires a string argument of length 5
*** aborting!
*** all protocols disconnected

(Not using SSL also gives same error)

The error seems to be coming from viff/runtime.py, line 320 from the struct.unpack routine. The length of string is 1 for a message, that gives the error.

I am able to run beginners test sometime, sometimes that also gives same error.

A similar issue was raised in Feb. on the list, but I didn't see any response to that (http://article.gmane.org/gmane.comp.cryptography.viff.devel/921).

If someone has any idea, please let me know.

Thanks in advance,

-Aseem.
Marcel Keller
2013-10-22 16:37:31 UTC
Permalink
Hi Aseem and Simone,

This is a known problem with later versions of Twisted. I recommend
using version 8 as this is successfully tested according to the
documentation. Berry Schoenmakers (***@win.tue.nl) told me that he has
fixed it with later versions of Twisted, but failed to send any patches.
In any case, I'm happy to include a patch solving the problem.

Best regards,
Marcel
Post by Aseem Rastogi
Mac OS 10.8
Seeding random generator with random seed 1945
Using SSL
Listening on port 4002
Will connect to <Player 3: localhost:4003>
I am Millionaire 2 and I am worth 83 millions.
*** bad data from Player 3: unpack requires a string argument of length 5
*** aborting!
*** all protocols disconnected
(Not using SSL also gives same error)
The error seems to be coming from viff/runtime.py, line 320 from the
struct.unpack routine. The length of string is 1 for a message, that
gives the error.
I am able to run beginners test sometime, sometimes that also gives same error.
A similar issue was raised in Feb. on the list, but I didn't see any
response to that
(http://article.gmane.org/gmane.comp.cryptography.viff.devel/921).
If someone has any idea, please let me know.
Thanks in advance,
-Aseem.
_______________________________________________
viff-devel mailing list (http://viff.dk/)
http://lists.viff.dk/listinfo.cgi/viff-devel-viff.dk
Fred Grossmann
2014-04-23 17:52:57 UTC
Permalink
Post by Marcel Keller
This is a known problem with later versions of Twisted. I recommend
using version 8 as this is successfully tested according to the
documentation. Berry Schoenmakers (berry <at> win.tue.nl) told me that he
has
Post by Marcel Keller
fixed it with later versions of Twisted, but failed to send any patches.
In any case, I'm happy to include a patch solving the problem.
Hey All,

it seems that viff expects the first packet to be an ID (with a length of
only one byte), but newer versions of twisted do not guarantee that this
will indeed be the first packet to be received. You can find a crudely fixed
version with working millionaires protocol at https://github.com/xony/viff

Cheers,
Fred
Marcel Keller
2014-05-01 15:18:06 UTC
Permalink
Post by Fred Grossmann
Post by Marcel Keller
This is a known problem with later versions of Twisted. I recommend
using version 8 as this is successfully tested according to the
documentation. Berry Schoenmakers (berry <at> win.tue.nl) told me that he
has
Post by Marcel Keller
fixed it with later versions of Twisted, but failed to send any patches.
In any case, I'm happy to include a patch solving the problem.
Hey All,
it seems that viff expects the first packet to be an ID (with a length of
only one byte), but newer versions of twisted do not guarantee that this
will indeed be the first packet to be received. You can find a crudely fixed
version with working millionaires protocol at https://github.com/xony/viff
Hi Fred,

Thanks for the fix, but it seems that the problem lies a bit deeper.
I've now submitted another fix to the repository.

Best regards,
Marcel

Loading...