> On Feb. 21, 2014, 9:01 p.m., Vinod Kone wrote: > > 3rdparty/libprocess/include/process/sequence.hpp, lines 75-88 > > <https://reviews.apache.org/r/17476/diff/4/?file=499673#file499673line75> > > > > Awesome graph. > > > > Maybe N1, N2 and N3 and T1 T2 and T3 would make it even better to read? > > > > Also consider s/T/F/ so that it is easy to understand that N => > > "next/notifier" and F ==> "future" in the code.
Agreed! s/T/F/. For the numbering part, I think the transition graph is already quite clear (from nothing -> 1 callback -> 2 callbacks), I am just too lazy to do that (you know I have to reformat:() :) > On Feb. 21, 2014, 9:01 p.m., Vinod Kone wrote: > > 3rdparty/libprocess/include/process/sequence.hpp, line 123 > > <https://reviews.apache.org/r/17476/diff/4/?file=499673#file499673line123> > > > > Why is this templated? Good catch! It was a template before, I forgot to change it back. > On Feb. 21, 2014, 9:01 p.m., Vinod Kone wrote: > > 3rdparty/libprocess/include/process/sequence.hpp, line 130 > > <https://reviews.apache.org/r/17476/diff/4/?file=499673#file499673line130> > > > > "__add" here is a bit confusing because this is not a continuation of > > "_add". Renamed to 'completed' and 'notified' to be more descriptive. - Jie ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/17476/#review35181 ----------------------------------------------------------- On Feb. 21, 2014, 9:02 p.m., Jie Yu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/17476/ > ----------------------------------------------------------- > > (Updated Feb. 21, 2014, 9:02 p.m.) > > > Review request for mesos and Benjamin Hindman. > > > Repository: mesos-git > > > Description > ------- > > See summary. > > One think I haven't done yet is to make it accepts '_Defer' types. (Just > curious why Future.onAny can accept defer without doing something special). > > Also, we can get rid of the SequencerProcess if we use a mutex to protect > 'last'. Not sure which one is better. > > The discard semantics here is a bit tricky. Basically, I wanna support > discarding a single callback without affecting other callbacks. Also, when > the Sequencer object is deleted, I wanna discard all the pending callbacks. > > > Diffs > ----- > > 3rdparty/libprocess/Makefile.am a7d199f > 3rdparty/libprocess/include/process/sequence.hpp PRE-CREATION > 3rdparty/libprocess/src/tests/sequence_tests.cpp PRE-CREATION > > Diff: https://reviews.apache.org/r/17476/diff/ > > > Testing > ------- > > make check > > repeated 1000 times > > > Thanks, > > Jie Yu > >
