> On Jan. 26, 2015, 2:16 p.m., Ben Mahler wrote:
> > Why is a base class needed here?

The idea here was to refactor RunState so that we can expose just the 
RunStateBase and not have to worry about exposing the Task and StatusUpdate 
protobufs from messages/messages.proto.  Doing this helps us prevent breakage 
with version upgrades when/if these internal definitions are modified.

The previous solution was to expose the entire slave/state.hpp file along with 
Task and StatusUpdate protobufs.  This would have worked fine, except that now 
successful upgrades on making sure that these internal definition don't change 
between mesos releases.

I understand that the upgrade can still fail if the definition of RunStateBase 
were to change, however, the scope is much narrower than the alternative 
approach.  Further, RunStateBase is needed only by containerizer/Isolator and 
so if one one to extend RunState for some other purpose, it would be just fine.


- Kapil


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/30237/#review69654
-----------------------------------------------------------


On Jan. 23, 2015, 9:38 p.m., Kapil Arya wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/30237/
> -----------------------------------------------------------
> 
> (Updated Jan. 23, 2015, 9:38 p.m.)
> 
> 
> Review request for mesos, Ian Downes, Niklas Nielsen, Till Toenshoff, and 
> Timothy Chen.
> 
> 
> Bugs: MESOS-2096
>     https://issues.apache.org/jira/browse/MESOS-2096
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> [4/10] Build Mesos Modules outside Mesos source tree.
> 
> This prepares RunStateBase to eventually be exposed in a public header
> include/mesos/slave/state.hpp.  This further prepares us to expose
> src/slave/containerizer/isolator.hpp as include/mesos/slave/isolator.hpp.
> 
> 
> Diffs
> -----
> 
>   src/slave/containerizer/isolator.hpp 
> a27c3e955ff25b87599af0ac2c97427a88b786f6 
>   src/slave/containerizer/isolator.cpp 
> 90a47c4fc8ce2d1a2aa09fa9ad905b76eab74c80 
>   src/slave/containerizer/isolators/cgroups/cpushare.hpp 
> 4ded0c4678ad61742a69e14233006a448f3fc3ca 
>   src/slave/containerizer/isolators/cgroups/cpushare.cpp 
> 90aabb85da452a3a508888f738379a73b7465dc6 
>   src/slave/containerizer/isolators/cgroups/mem.hpp 
> 2fa755571b4d21b8b13301fcfd57ae05ea66e6e6 
>   src/slave/containerizer/isolators/cgroups/mem.cpp 
> 711d66d7771cac13be831d73af3ef570d6785473 
>   src/slave/containerizer/isolators/cgroups/perf_event.hpp 
> e511c3efe4cc4ec13cc74bdcda993477a4db2e36 
>   src/slave/containerizer/isolators/cgroups/perf_event.cpp 
> 6f67164d3963769148fb1749d1d590b7b2999fdb 
>   src/slave/containerizer/isolators/filesystem/shared.hpp 
> 727f63e16fa69293c472a4b6b95fb0be04e689c8 
>   src/slave/containerizer/isolators/filesystem/shared.cpp 
> 5c347af803bf512698dd580d6533c0103b289427 
>   src/slave/containerizer/isolators/namespaces/pid.hpp 
> 8da6cccff04edf11fa37f5a33648821299cf49d5 
>   src/slave/containerizer/isolators/namespaces/pid.cpp 
> fdd430e946c386e37ffc5b24907063180cddac17 
>   src/slave/containerizer/isolators/network/port_mapping.hpp 
> f1e23526bc18b23f187c5d38e63d366cb509fff0 
>   src/slave/containerizer/isolators/network/port_mapping.cpp 
> 2d813369bd844ad743f7d3f18dbab29b2467f2c8 
>   src/slave/containerizer/isolators/posix.hpp 
> 7a667e3e30712b98484947520b58965d3b768659 
>   src/slave/containerizer/isolators/posix/disk.hpp 
> cf4e143eb50cf5ee59ca0d47dd6f8d0a845c3732 
>   src/slave/containerizer/isolators/posix/disk.cpp 
> bb1779fd644a33527068868a45cf9c19d7732727 
>   src/slave/containerizer/mesos/containerizer.cpp 
> d712278428889ebdfd598537690138329d8464f0 
>   src/slave/state.hpp 70777cf6ab681c29ca4df601fe47903e1dbdf41f 
>   src/tests/isolator.hpp 2c3a694fade996df6fb679e78f96876a7d32d214 
> 
> Diff: https://reviews.apache.org/r/30237/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Kapil Arya
> 
>

Reply via email to