Re: [Qemu-devel] [PATCH v2 2/2] full introspection support for QMP

[Amos Kong]

On Fri, Dec 20, 2013 at 07:03:57PM +0100, Paolo Bonzini wrote:
> Il 20/12/2013 12:57, Amos Kong ha scritto:
> > { 'type': 'DataObjectBase',
> >   'data': { '*name': 'str', 'type': 'str' } }
> > { 'union': 'DataObjectMemberType',
> >   'discriminator': {},
> >   'data': { 'reference': 'str',
> >             'undefined': 'DataObject',
> >             'extend': 'DataObject' } }
> 
> What is the purpose of "undefined"?  I don't see any occurrence of any
> of "undefined" or "extend" in the sample.


| { 'type': 'VersionInfo',
|   'data': {'qemu': {'major': 'int', 'minor': 'int', 'micro': 'int'},
|           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

| {
|   "name": "VersionInfo", 
|   "type": "type", 
|   "data": [
|~      {
|~          "name": "qemu", 
|~          "optional": false, 
|~          "type": {
|~              "type": "undefined-struct", 
|~              "data": [
|~                  {
|~                      "name": "micro", 
|~                      "optional": false, 
|~                      "recursive": false, 
|~                      "type": "int"
|~                  }, 
|~                  {
|~                      "name": "minor", 
|~                      "optional": false, 
|~                      "recursive": false, 
|~                      "type": "int"
|~                  }, 
|~                  {
|~                      "name": "major", 
|~                      "optional": false, 
|~                      "recursive": false, 
|~                      "type": "int"
|~                  }
|~              ]
|~          }
|~      }, 

> > { 'type': 'DataObjectMember',
> >   'data': { 'type': 'DataObjectMemberType', '*name': 'str',
> >             '*optional': 'bool', '*recursive': 'bool' } }
> > { 'type': 'DataObjectCommand',
> >   'data': { '*data': [ 'DataObjectMember' ],
> >             '*returns': 'DataObject',
> >             '*gen': 'bool' } }
> > { 'type': 'DataObjectEnumeration',
> >   'data': { 'data': [ 'str' ] } }
> > { 'type': 'DataObjectType',
> >   'data': { 'data': [ 'DataObjectMember' ] } }
> > { 'type': 'DataObjectUndefinedStruct',
> 
> Perhaps Unnamed or Anonymous?
 
Anonymous is good.

> >   'data': { 'data': [ 'DataObjectMember' ] } }
> > { 'type': 'DataObjectUnion',
> >   'data': { 'data': [ 'DataObjectMember' ], '*base': 'str',
> >             '*discriminator': 'str' } }
> > { 'union': 'DataObject',
> >   'base': 'DataObjectBase',
> >   'discriminator': 'type',
> >   'data': {
> >     'command': 'DataObjectCommand',
> >     'enumeration': 'DataObjectEnumeration',
> >     'type': 'DataObjectType',
> >     'undefined-struct': 'DataObjectUndefinedStruct',
> >     'reference-type': 'String',
> >     'unionobj': 'DataObjectUnion' } }
> > { 'command': 'query-qmp-schema', 'returns': ['DataObject'] }
> 
> I think forcing expansion of everything that isn't unnamed/anonymous
> makes the schema much larger and unwieldy.  Otherwise looks great!

We want to provide more useful metadata, and used some enum/unions to
indicate the dynamic type.

In the output, some simple data is processed too unwieldy. In another
side, some complex data is described clearly. It's also caused by some
limitation of QAPI infrastructure.
 
> Paolo

-- 
                        Amos.