----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/41168/#review111427 -----------------------------------------------------------
src/tests/mesos.hpp (line 152) <https://reviews.apache.org/r/41168/#comment171604> Then why not return this as Owned? Capturing this requirement as a comment is brittle and prone to bugs. Also, it's clear you've got a global invariant that you're trying to comment on here, in particular, that when someone calls `CreateContainerLogger` that container logger will get passed into a (some? which?) containerizer which will be responsible for cleaning it up! But as a generic `MesosTest` function why can I not just call `CreateContainerLogger` at any point and _not_ pass it into a containerzer? Then does this comment still apply? While we could just clean up the comment (i.e., "the caller owns the returned pointer and must clean it up"), by returning an `Owned` no comment is necessary! Note that the one place where we make exceptions to this rule is factory functions, like `ContainerLogger::create`, but even there I'd love to see a world where we use `Owned`. - Benjamin Hindman On Dec. 15, 2015, 8:40 p.m., Joseph Wu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/41168/ > ----------------------------------------------------------- > > (Updated Dec. 15, 2015, 8:40 p.m.) > > > Review request for mesos, Benjamin Hindman and Artem Harutyunyan. > > > Bugs: MESOS-4088 > https://issues.apache.org/jira/browse/MESOS-4088 > > > Repository: mesos > > > Description > ------- > > Adds a helper that creates and initializes the `ContainerLogger`. This is > used in some tests that use test-only containerizer constructors. > > > Diffs > ----- > > src/tests/mesos.hpp 689efb40dc22337766b624218d2504d5de3e5cd8 > src/tests/mesos.cpp a501f04dfbf9c93c2196e83b512519e925a67c97 > > Diff: https://reviews.apache.org/r/41168/diff/ > > > Testing > ------- > > This will be tested later in the review chain. > > > Thanks, > > Joseph Wu > >
