> On June 28, 2016, 2:50 p.m., Alexander Rojas wrote: > > include/mesos/authorizer/authorizer.proto, lines 90-91 > > <https://reviews.apache.org/r/49319/diff/1/?file=1431917#file1431917line90> > > > > This comment is rather incomplete: > > > > 1. Either they are both set or none. > > 2. When none are set it is because the task has not yet created the > > directory (because is scheduled) or it is finished and there was a lag > > between access and destruction of the enpoint. > > > > All in all, we should mention that having none set is an exceptional > > case, and could be treated by returning an error. > > Joerg Schad wrote: > 1. Are you sure from the setting code that 1. is true? > 2. I am not sure this needs to mentioned here as this is just describes > the semantic, and a module writer has to deal with that no matter how > frequent it is. > > We can surely mention this is an exceptional case, but I would answer > similarly as for 2. > > Alexander Rojas wrote: > 1. Yes, It is true! > 2. if there is no object to authorize too, how do a module writer has to > react to it? This is an exceptional case and needs to be mention that it does > occur > > Joerg Schad wrote: > How about: > // This action will have an object which might either have both a > `FrameworkInfo`, > // and `ExecutorInfo`, or in exceptional cases nothing set. > > Joerg Schad wrote: > Btw Maybe we should later add a check which verfies this invariant...
Are you saying we should check for either "both" vs. "none" to make sure we never run into "some"; as in `FrameworkInfo` being set but `ExecutorInfo` not being set? - Till ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/49319/#review139796 ----------------------------------------------------------- On June 28, 2016, 2:35 p.m., Joerg Schad wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/49319/ > ----------------------------------------------------------- > > (Updated June 28, 2016, 2:35 p.m.) > > > Review request for mesos, Adam B and Alexander Rojas. > > > Bugs: MESOS-5730 > https://issues.apache.org/jira/browse/MESOS-5730 > > > Repository: mesos > > > Description > ------- > > The current semantic is that these fields might not be set. > > > Diffs > ----- > > include/mesos/authorizer/authorizer.proto > fc76796022a6fa3d36a1447c476980868d42c2d0 > > Diff: https://reviews.apache.org/r/49319/diff/ > > > Testing > ------- > > > Thanks, > > Joerg Schad > >