Tor controller strategy: recognizing log messages

[Krishna Sankar]

At least there are two ways to go - a Dbus model to capture the events or harvesting the log. As you accurately pointed out, the ones that appear are the ones we do not know about. 

So, parsing the events from the logs is a good strategy. I think we can write this in a py-based rules engine of some sort. Don't know if it is feasible to have a unique event code for this.

Cheers
<k/>

> -----Original Message-----
> From: owner-or-dev at freehaven.net 
> [mailto:owner-or-dev at freehaven.net] On Behalf Of Roger Dingledine
> Sent: Thursday, July 13, 2006 9:37 PM
> To: or-dev at freehaven.net
> Subject: Tor controller strategy: recognizing log messages
> 
> Hi folks,
> 
> (I am mostly thinking of Nick and Matt Edman as the 
> recipients of this, but I figured or-dev was a fine spot for 
> it in case one of you knows the right answer.)
> 
> Vidalia could become smarter if it read Tor's log messages 
> and interpreted them better for the user.  ("I know why Tor 
> is giving this error -- I recommend you foo.") The step after 
> this would be to automatically do something smart when a 
> given event happens.
> 
> But the best way to recognize these events right now is to 
> read Tor's log messages and recognize certain strings.
> 
> Is this an ok long-term strategy? I don't really want to 
> write a new protocol to inform the controller of any event 
> that it might possibly want to know about (since a good chunk 
> of the relevant events are the ones we don't realize will be 
> issues). And I sure don't want to standardize what log 
> messages are given in what situations. But I also want the 
> flexibility to make Tor's log messages better down the road.
> 
> Is there an easy answer?
> 
> --Roger
>