----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/26476/#review56846 -----------------------------------------------------------
A few higher level questions: (1) What motivated this? Concretely, which performance aspect of Mesos is this improving? In the past, we eliminated copies of our Option,Try,Future,Result family because we found that the copying of large Option,Try,Future,Result<Registry> types was degrading performance. In general there should be a well understood benefit, especially when we're increasing the esotericism of the code in the name of performance. :) (2) When is this better? When is this worse? It looks like None() options are now more expensive? Did you measure Option performance with any benchmarks? (3) Curious why you introduce the new public reset() method, since most callers use `option = None()`. Would be great to avoid introducing another way to do it. - Ben Mahler On Oct. 15, 2014, 9:29 p.m., Joris Van Remoortere wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/26476/ > ----------------------------------------------------------- > > (Updated Oct. 15, 2014, 9:29 p.m.) > > > Review request for mesos, Benjamin Hindman and Niklas Nielsen. > > > Repository: mesos-git > > > Description > ------- > > Remove dynamic allocations from Option class. > > > Diffs > ----- > > 3rdparty/libprocess/3rdparty/stout/include/stout/option.hpp 47fe92c > > Diff: https://reviews.apache.org/r/26476/diff/ > > > Testing > ------- > > make check > support/mesos-style.py > valgrind (reduced allocation count) > > > Thanks, > > Joris Van Remoortere > >