Since we're investigating ways to reduce power consumption, it would be useful to see what effect reader mode has. To detect if a page is readable, we: 1) Fire a web worker 2) Send a string of the entire HTML document to that worker 3) Parse the string into a DOM(ish) representation using JSDOMParser 4) Perform a number of complex algorithms on this result We go through these steps for every page visited, so it's possible that reader mode could be consuming a non-trivial amount of power.

Things have changed since this bug was filed, and (luckily!) we no longer do all this work on each page load. We only fire up a web worker and do the parser when the user taps the reader view button. We only call the `isProbablyReaderable` function on each page load: http://mxr.mozilla.org/mozilla-central/source/toolkit/components/reader/ReaderMode.jsm?force=1#120

Margaret, are confident that the cost is small enough that we can close this bug? (I don't mind either way and am happy to defer to your decision.)

(In reply to Nicholas Nethercote [:njn] from comment #2) > Margaret, are confident that the cost is small enough that we can close this > bug? (I don't mind either way and am happy to defer to your decision.) I tried digging into our telemetry data to see how often users try loading pages in reader view, and I couldn't find a probe for that, so I filed bug 1204896. I imagine that people load pages in reader view a lot less often that they load normal pages, so there are probably other power consumption bugs that would be higher priority. I also don't know what we would do to reduce the cost of parsing a page when a user requests it. I think bnicholson filed this bug because we used to do this on every page load, so now that that's not the case, I'll just close this out.