----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/41474/#review111287 -----------------------------------------------------------
3rdparty/libprocess/README.md (line 106) <https://reviews.apache.org/r/41474/#comment171398> Per discussion, adjust this text to give more context: * all callbacks are invoked synchronously in the process that satisfies the future * sometimes this is bad * you can use defer() to arrange for an operation to be performed asynchronously in the context of a specified process Then fix up comments in code sample and discussion below. 3rdparty/libprocess/README.md (line 109) <https://reviews.apache.org/r/41474/#comment171382> What is "the specified process"? 3rdparty/libprocess/README.md (line 111) <https://reviews.apache.org/r/41474/#comment171385> "with it" or "on it"? (I'm not actually sure.) 3rdparty/libprocess/README.md (line 122) <https://reviews.apache.org/r/41474/#comment171389> "be invoked" 3rdparty/libprocess/README.md (line 130) <https://reviews.apache.org/r/41474/#comment171390> "be invoked" 3rdparty/libprocess/README.md (line 133) <https://reviews.apache.org/r/41474/#comment171388> Avoid using "runs" twice in the same sentence. 3rdparty/libprocess/README.md (line 150) <https://reviews.apache.org/r/41474/#comment171405> The need to use `lambda::_1` here is a bit unfortunate. 3rdparty/libprocess/README.md (line 156) <https://reviews.apache.org/r/41474/#comment171399> "the one in which it is registered" or "the process that satisfies the future"? 3rdparty/libprocess/README.md (line 157) <https://reviews.apache.org/r/41474/#comment171400> `self()` is not used in the code sample. 3rdparty/libprocess/README.md (line 158) <https://reviews.apache.org/r/41474/#comment171401> There are two callbacks in the code sample above, not clear which one you're referring to. 3rdparty/libprocess/README.md (line 224) <https://reviews.apache.org/r/41474/#comment171406> I don't think we should be encouraging people to write code when they are "in doubt" :) There have been non-deterministic bugs that have arisen from missing `defer`s -- i.e., whether a lack of `defer` causes problems can depend on the vagaries of the libprocess scheduler. e.g., https://reviews.apache.org/r/40610/diff/1#index_header -- would it be feasible/useful to walk through a simple example? - Neil Conway On Dec. 16, 2015, 11:46 p.m., Greg Mann wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/41474/ > ----------------------------------------------------------- > > (Updated Dec. 16, 2015, 11:46 p.m.) > > > Review request for mesos, Benjamin Hindman, Ben Mahler, Joris Van Remoortere, > and Neil Conway. > > > Bugs: MESOS-3996 > https://issues.apache.org/jira/browse/MESOS-3996 > > > Repository: mesos > > > Description > ------- > > Added documentation for `defer` in libprocess README. > > > Diffs > ----- > > 3rdparty/libprocess/README.md c3f309a7dc1c94882c4cc97eeaf0736c2fca0ba5 > > Diff: https://reviews.apache.org/r/41474/diff/ > > > Testing > ------- > > Viewed the markdown in a branch on GitHub: > https://github.com/mesosphere/mesos/tree/defer_docs/3rdparty/libprocess > > > Thanks, > > Greg Mann > >