Re: [Qemu-devel] [PATCH 22/47] block: make device optional in BlockInfo

[Kevin Wolf]

Am 24.07.2012 13:04, schrieb Paolo Bonzini:
> Targets of a mirroring operation will not have a device.  Once we have
> -blockdev or equivalent, "detached" block devices and non-anonymous
> backing files also will not have a device.
> 
> Signed-off-by: Paolo Bonzini <address@hidden>
> ---
>  qapi-schema.json |    5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/qapi-schema.json b/qapi-schema.json
> index fca1806..b00d8c6 100644
> --- a/qapi-schema.json
> +++ b/qapi-schema.json
> @@ -443,7 +443,8 @@
>  # Block device information.  This structure describes a virtual device and
>  # the backing device associated with it.
>  #
> -# @device: The device name associated with the virtual device.
> +# @device: #optional The device name associated with the virtual device.
> +#          Always included in the output of query-block.
>  #
>  # @type: This field is returned only for compatibility reasons, it should
>  #        not be used (always returns 'unknown')
> @@ -465,7 +466,7 @@
>  # Since:  0.14.0
>  ##
>  { 'type': 'BlockInfo',
> -  'data': {'device': 'str', 'type': 'str', 'removable': 'bool',
> +  'data': {'*device': 'str', 'type': 'str', 'removable': 'bool',
>             'locked': 'bool', '*inserted': 'BlockDeviceInfo',
>             '*tray_open': 'bool', '*io-status': 'BlockDeviceIoStatus'} }

Is this really a compatible change? That 'device' is basically the
unique key by which block device are identified doesn't exactly make
feel more comfortable about the change.

Of course, not making it optional means that basically we need to go the
way of referencing the block device in query-block-jobs immediately
instead of thinking about it later. You know that I preferred this from
the start, and this change is just another detail that makes me think
it's the right thing to do.

Kevin