-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/27555/
-----------------------------------------------------------
(Updated Nov. 4, 2014, 10:01 p.m.)
Review request for mesos, Ben Mahler and Vinod Kone.
Changes
-------
BenM's suggestion on the 'remove' semantics.
Bugs: MESOS-1974
https://issues.apache.org/jira/browse/MESOS-1974
Repository: mesos-git
Description
-------
The purpose of the refactor is to support persistent disk resources.
Here are the main things I've done in this refactor:
1) Resource objects in Resources are stored in minimal format
(validated/non-zero). That allows us to kill isAllocatable, allocatable,
isZero, etc.
2) 'matches' needs to be split into two pieces: one for combining and one for
removing, in order to support persitent disk resource. For example, one cannot
combine two Resource object with DiskInfo (it's like two disks), however, you
can do removal if they are identical.
3) Some of the interfaces are not intuitive (e.g., <=, see details in the
ticket). I removed them in favor of more explicit interfaces.
4) Unified all the validation code.
5) Adjusted the tests accordingly.
Diffs (updated)
-----
include/mesos/resources.hpp 0e37170
src/cli/execute.cpp ddaa20d
src/common/resources.cpp e9a0c85
src/examples/low_level_scheduler_libprocess.cpp 7ef5ea7
src/examples/low_level_scheduler_pthread.cpp 6e233a1
src/examples/no_executor_framework.cpp f98a073
src/examples/test_framework.cpp 187a611
src/master/drf_sorter.cpp 5464900
src/master/hierarchical_allocator_process.hpp 31dfb2c
src/master/http.cpp 3189933
src/master/master.cpp d914786
src/tests/allocator_tests.cpp 58e15aa
src/tests/gc_tests.cpp f7747e2
src/tests/master_tests.cpp 2e52574
src/tests/mesos.hpp c1d64a7
src/tests/resource_offers_tests.cpp fe66432
src/tests/resources_tests.cpp 3e50889
src/tests/slave_recovery_tests.cpp 98e059f
src/tests/sorter_tests.cpp 0516ab5
Diff: https://reviews.apache.org/r/27555/diff/
Testing
-------
make check
Thanks,
Jie Yu
