> On Sept. 22, 2016, 7:58 p.m., Joseph Wu wrote: > > 3rdparty/libprocess/src/openssl.cpp, line 493 > > <https://reviews.apache.org/r/52153/diff/1/?file=1508015#file1508015line493> > > > > I recall that we prefer `.at(...)` even when there's an existence check > > directly preceeding the map access. > > Benjamin Bannier wrote: > I was briefly wondering why that would be prefered (terminate if key does > not exist?), but `map::at` being `const` sold me.
Ensuring const access is why I've been doing it, although it seems a bit worrying that `map::at` will throw an exception. If an exception is thrown within a `Process` execution context, libprocess will catch it and terminate the `Process`. After this point, dispatches to the terminated `Process` will be dropped, and we will only bail if it's the `Process` that main thread is waiting for. Ideally we guarantee that we will crash if a Process throws an exception. - Benjamin ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/52153/#review150062 ----------------------------------------------------------- On Sept. 22, 2016, 9:43 a.m., Till Toenshoff wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/52153/ > ----------------------------------------------------------- > > (Updated Sept. 22, 2016, 9:43 a.m.) > > > Review request for mesos, Benjamin Bannier, Joris Van Remoortere, and Joseph > Wu. > > > Bugs: MESOS-5320 > https://issues.apache.org/jira/browse/MESOS-5320 > > > Repository: mesos > > > Description > ------- > > see summary. > > > Diffs > ----- > > 3rdparty/libprocess/src/openssl.cpp > c09cdc89509e4e4ca4c8a0f4fb0a57156a3a6091 > > Diff: https://reviews.apache.org/r/52153/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Till Toenshoff > >
