Could you look into this ancient pull (#4871) and perhaps check or base this one on mine? Also I'd like to get a more detailed description of what your patch does... I didn't quite understand everything.

Any reason why you're not just using fTorNoProxy in LookupIntern?

@sipa Yea, that breaks unrelated things (eg you end up not sending the right version message as the *.onion is now mapped to a (probably local) IP instead). Also, you would break getpeerinfo.

@Diapolo Hmm? It looks like any merge conflicts with that should be cosmetic in nature (ie take no effort to fix). I'd rather keep them separate.

@TheBlueMatt My intention was to get some review and perhaps discover changed behaviour of my pull, as it seems no one tested it (and I'm missing a testplan yes...).

@Diapolo at least the issue is the same here: we don't have any proxy tests

Can you convince me that this is really needed? I see it being a pain to maintain and keep secure (e.g. avoid leaks) esp if in the future we get better integration with tor for hidden services, like being able to provide HS keys for dos mitigation.

NACK. I agree with @gmaxwell and think this complicates the code for a feature I'd expect no one but @TheBlueMatt will use. No tests are provided either, so it could break any time.

Tests should be written, agreed, but this is incredibly useful for anyone using a tor middlebox. There are several of those around these days, and its not that much code to support a case like that.

Closing due to the changing-exit-every-time change, as this would be inherintly less anonymous than using the proxy exlicitly thereafter.