Well, no rants, but I'm in qualified agreement with Scott [just this once, heh]... that yes, those of us stuck in 80x25 terminals and antique text comment databases could use a multiline format. It the project is concerned about the replace vs. add semantics, one could add two new exclude[exit]nodesmethod options. Where method is replace or add. Yet it is just an ease of use thing, and does have a certain impact on downstream controller code. So as long as the config does what the docs say it does, in whatever mode it ends up taking, that's fine, people will hack and make do either way. Config file and controller interface should act the same though.
Also, regarding the interaction with HS directory lookups and excludenodes... i would suggest that specification in excludenodes should prevent all contact with such node for all reasons. Or just make another option for how to handle that case as well. This is more important than the above paragraph. As one could have a node that is a 'bad' exit through no fault/intent of its operator... such as being plugged into a non-ideal isp... yet it would still be perfectly useful when acting as a non-exit or directory provider. *********************************************************************** To unsubscribe, send an e-mail to majord...@torproject.org with unsubscribe or-talk in the body. http://archives.seul.org/or/talk/