[tor-dev] tor's definition of 'median'

[Nick Mathewson]

On Wed, Aug 12, 2015 at 5:34 PM, nusenu <nusenu at openmailbox.org> wrote:
> from today's measurement meeting:
>
>> 15:00:20 <virgil> karsten: I've decided I'm going to fix the definition of median
>> 15:00:26 <virgil> in the tor sourcecode
>> 15:00:36 <karsten> virgil: is it broken?
>> 15:00:53 <karsten> or just not specified as clearly as it should be?
>> 15:01:01 <virgil> for ordered list {a,b,c,d}, it returns b instead of (b+c)/2.
>> 15:01:24 <karsten> yes. maybe that's for a reason (which I don't know).
>> 15:01:40 <virgil> I look forward to hearing this reason when my patch is rejected.
>> 15:01:41 <karsten> like, using value (b+c)/2 would break for some reason, whereas any of a, b, c, d would be fine.
>> 15:01:45 <Sebastian> you cannot do that
>> 15:01:51 <Sebastian> without breaking Tor's voting
>> 15:02:21 <Sebastian> Tor's specification requires low median for a bunch of directory stuff
>
>
> I'd be interested in the reason as well.

The correct fix here is to update the code documentation to define the
functions as returning the low-median, and to update dir-spec.txt to
say so too.  I'd accept documentation patches like that.

Changing the code to return the mean of the two center elements from
even arrays would break all authority voting, and wouldn't actually be
useful.

-- 
Nick