Re: [Qemu-devel] [PATCH RFC 3/3] tests: add test cases for qapi event support

[Wenchao Xia <xiawenc@linux.vnet.ibm.com>]

于 2013/12/16 10:55, Wenchao Xia 写道:
> 于 2013/12/13 21:36, Eric Blake 写道:
>> On 11/12/2013 06:44 PM, Wenchao Xia wrote:
>>> Signed-off-by: Wenchao Xia <xiawenc@linux.vnet.ibm.com>
>>> ---
>>
>>> +++ b/tests/qapi-schema/qapi-schema-test.json
>>> @@ -93,3 +93,15 @@
>>>       '*u16' : [ 'uint16' ],
>>>       '*i64x':   'int'     ,
>>>       '*u64x':   'uint64'  } }
>>> +
>>> +# testing event
>>> +{ 'type': 'EventStructOne',
>>> +  'data': { 'struct1': 'UserDefOne', 'string': 'str', '*enum2':
>>> 'EnumOne' } }
>>> +
>>> +{ 'event': 'EVENT_A' }
>>> +{ 'event': 'EVENT_B',
>>> +  'data': { } }
>>
>> Do we really have events with no associated information (the mere name
>> of the event carrying the full data?)
>>
>
>    I guess EVENT_SHUDDOWN is one example, that have no associated info,
> except time stamp.
>

   I think you questioned why EVENT_B is needed? I added it before to
allow that, both define can work.

>
>>> +{ 'event': 'EVENT_C',
>>> +  'data': { '*a': 'int', '*b': 'UserDefOne', 'c': 'str' } }
>>> +{ 'event': 'EVENT_D',
>>> +  'data': { 'a' : 'EventStructOne', 'b' : 'str', '*c': 'str',
>>> '*enum3': 'EnumOne' } }
>>
>> Looks like reasonable support for encoding existing events.
>>
>>> + OrderedDict([('event', 'EVENT_A')]),
>>> + OrderedDict([('event', 'EVENT_B'), ('data', OrderedDict())]),
>>
>> Shouldn't the omission of 'data' be the same as 'data':{}?
>>
>
>    I didn't touch the "{ }" parsing part, maybe an existing parsing
> style, will check it.
>
>>> +++ b/tests/test-qmp-event.c
>>> @@ -0,0 +1,250 @@
>>> +/*
>>> + * QMP Input Visitor unit-tests.
>>> + *
>>> + * Copyright (C) 2011 Red Hat Inc.
>>
>> Unusual choice of copyright year and ownership.
>>
>   yep, will fix.
>
>