[tor-dev] Stem code review 2012-12-21
atagar at torproject.org
Sat Dec 22 22:19:17 UTC 2012
Hi Sean, thanks for the code review!
> Would GETINFO/GETCONF cache hit logging benefit from log_once? Or even something between log and log_once (e.g. rate_limited_log)?
Interesting thought, but I don't think so. When I look for cache hits
it's usually to answer the question of 'did doing activity X result in
requesting parameter Y'. Having stem log once or drop logging when it
exceeds a certain rate would make that more confusing.
> Is there any particular reason for moving away from readthedocs.org? I'm curious if there was more than a desire to self-host as much as possible.
I talked a bit about the reason for moving on
'https://trac.torproject.org/7324'. Mostly the reasons were (in order
by how much I cared about them)...
1. A shorter and more official sounding tpo domain.
2. The daily autobuilds on read-the-docs were always stuck, so the
documentation was only updated when I manually built it. This meant
that we usually had stale docs. Now that we're on our own hardware I
also bumped the rate of checking for new content so the site updates
within five minutes of a push.
3. Self-hosting provides better security and visitor privacy. I didn't
experiment to see how good the sandboxing of read-the-docs builds
were, but a community system that runs arbitrary code is a little
> The heartbeat time tracking is a good idea. But, I wonder, should this bother with an accessor method? Could this work as just well as an (non-callable) attribute?
I'd be fine with that. The only advantage I can think for having it as
a method is if we want custom handling in certain situations (for
instance, returning None if we're disconnected). However, we don't
presently do that so I could go either way.
More information about the tor-dev