UENISHI Kota

there's no include in msgpack-idl even its in the msgpack-idl spec.

FP16 support

1

At least we know with FP16 model Communicator's `bcast_data` does not work. ```diff diff --git a/tests/chainermn_tests/communicator_tests/test_communicator.py b/tests/chainermn_tests/communicator_tests/test_communicator.py index a0ff350..f03fa5d 100644 --- a/tests/chainermn_tests/communicator_tests/test_communicator.py +++ b/tests/chainermn_tests/communicator_tests/test_communicator.py @@ -242,6 +242,12 @@ def test_communicator_cpu(param):...

Port Chainer#4191 or use Chainer's BN implementation

2

ChainerMN has mostly-copied BatchNormalization code (but several AllReduce added), which means potential bugs from Chainer could also be imported. https://github.com/chainer/chainer/pull/4191 could be one of them; porting it to ChainerMN seems...

As a backlog that came from #234 and #237 .

HDDS-6483. Change delete key on OMKey(s)Delete

3

## What changes were proposed in this pull request? This patch changes: - The key format of delete table from full key to `-` - OzoneClient protocol of delete key...

Just adding to `webmachine_error.erl` will work for CS, but even without this CS just works. - https://github.com/basho/riak_cs/pull/1118 - http://tools.ietf.org/html/rfc6585

Log compression [JIRA: RIAK-1590]

4

Logs generated by log handler could be compressed and named like `access_log.blah.log.gz`. Originated from customer question: zd://10361.

Snapshot consistency discussion revival

2

Looking around the snapshot code, I just found that they were almost ported from Chainer. That said, discussion in https://github.com/chainer/chainer/issues/6763 also applies to PPE. Just for heads up...