Some omitempty usage modifications

[zhouhao3]

Optional fields should be added omitempty
Signed-off-by: zhouhao zhouhao@cn.fujitsu.com

[wking]

“omitepty” → “omitempty”.

[wking]

Besides the typos, 5a318bf looks good to me.

[philips]

LGTM

[vbatts]

LGTM

[stevvooe]

omitempty has nothing to do with REQUIRED vs OPTIONAL.

Please revert this change.

[zhouhao3]

@stevvooe In addition, after reading your discussion, I am not very clear when omitempty should be used, that is, why do you think this change is wrong, I did not see the specific provisions.

[stevvooe]

omitempty influences when Go json package emits values for zero-valued fields. Please read https://golang.org/pkg/encoding/json/. It should not be used to try to comply with the required and optional provisions of the specification, as simply doesn't do that.

Typically, omitempty should be used on every field unless you want it to emit a zero-value. An example might be if you want to have an empty slice be emitted, rather than omitting the field. I don't think there are any cases in OCI where it make sense to not have omitempty.

[zhouhao3]

@stevvooe So. I can't revert it，because the content have changed since it was merged. Would I need to reopen a PR to restore it?

[wking]

On Mon, Feb 06, 2017 at 07:43:01PM -0800, Stephen Day wrote: Typically, `omitempty` should be used on every field unless you want it to emit a zero-value. An example might be if you want to have an empty slice be emitted, rather than omitting the field. I don't think there are any cases in OCI where it make sense to not have `omitempty`.

The manifest-list ‘manifests’ entry is currently explicitly REQUIRED and it explicitly allows a zero-length array [1]. I'm not sure why we don't just make it OPTIONAL (less work for the same amount of required information), but as the spec stands it would not be a good idea to make ManifestList.Manifests omitempty (as I pointed out in [2]). [1]: https://github.com/opencontainers/image-spec/blame/v1.0.0-rc4/manifest-list.md#L23-L24 [2]: #557 (comment)