MorganS:
Yes, I might have something useful to say but without seeing what else has been posted, it's difficult to say.
People didn't provide much detail. Essentially, the idea they presented was to directly attached the rotary encoder to a secondary measurement spindle that has the line wrapped around it a single time.
My concern with the rotary encoder is that it may experience mechanical wear if it's constantly being turned, though I don't really know what the lifetimes are on these things. I'm also curious if a rotary encoder would have trouble reading accurately at higher rotation rates. I'm not very familiar with their accuracy, but I would imagine that an optical sensor would be able to deal with high rates of rotation more easily than a rotary encoder. I could very well be wrong about that though.