On 22 November 2016 at 14:56, Rui Ueyama <ruiu at google.com> wrote: > OK. I cannot think of a reason to stick with 0x100000. I'll give it some time for other comments since changing this will cause a lot of tests to churn.