I am attempting to use Chutney to test some hidden service related functionality in Tor. Specifically distributed hidden services involving more than one node.
Furthermore, to extensively test aspects of this, the network needs to change. At the moment, I have just looked at nodes coming up and down, but this could probably be extended also.
Currently I believe the support for this functionality is limited to a patch (very functional, I have been using it successfully) that allows nodes to start after a delay [1].
While this does work, this could make any tests that use delays respond differently depending on what system they are ran.
The best solution to this that I can see is using some event based system, such that chutney (using an existing Tor controller) connects to the nodes as necessary, to check their state, and trigger any events as necessary.
A few high level examples that I would like to use: - when node 010h has uploaded a descriptor for service <service>, start node 011h - when any of nodes (000a, 001a, 002a) have a service descriptor for <service>, start node 011h
Do others using/developing chutney think this would be a useful addition, and is anyone already working on doing this or similar?
On Tue, Nov 26, 2013 at 1:32 PM, Christopher Baines cbaines8@gmail.com wrote: [...]
Do others using/developing chutney think this would be a useful addition, and is anyone already working on doing this or similar?
I think some scripting facility of this kind could be of use.
I also think that Chutney needs a pretty big refactoring, perhaps based on stem. The code is pretty ugly and hard to follow as it is, IMO, and making it more powerful will only make it worse.
yrs,