[Qemu-devel] Re: qemu-img help missing backing file

[Kevin Wolf]

Am 30.03.2010 15:04, schrieb Juan Quintela:
> Kevin Wolf <address@hidden> wrote:
>> Am 30.03.2010 14:32, schrieb Alexander Graf:
>>>
>>> On 30.03.2010, at 14:30, Kevin Wolf wrote:
>>>
>>>> Am 30.03.2010 14:22, schrieb Alexander Graf:
>>>>> Howdy,
>>>>>
>>>>> I just wanted to create a backed qcow2 image and was irritated by 
>>>>> qemu-img not showing me the correct command line option. It's just 
>>>>> missing from the list:
>>>> [...]
>>>>> Is this intentional? The actual command still works:
>>>>>
>>>>>> address@hidden:~/git/qemu> qemu-img-kvm create -f qcow2 -b 
>>>>>> /media/studio/images/SUSE/s390/sles11.raw /dev/shm/sles11-zipl.qcow2
>>>>>> Formatting '/dev/shm/sles11-zipl.qcow2', fmt=qcow2 size=21474836480 
>>>>>> backing_file='/media/studio/images/SUSE/s390/sles11.raw' encryption=off 
>>>>>> cluster_size=0 
>>>>
>>>> -b still works to maintain compatibility with older versions. The
>>>> documented way is -o backing_file=foo (the qemu-img output you quote
>>>> even contains this option).
>>>
>>> Interesting O_o. Maybe it'd be a good idea to give some examples in the 
>>> help output? The same way the -e and -b options were having examples there 
>>> too. That really makes them easier to find.
>>
>> Feel free to submit a patch. ;-)
>>
>> Examples sound like an easy way to make these features more visible
>> again. Even better, but a bit more work, would be to include a
>> dynamically generated list of all supported options like this:
>>
>> backing_file: File name of a base image (qcow, qcow2, vmdk)
>> backing_fmt:  Image format of the base image (qcow2)
>> encryption:   Encrypt the image (qcow, qcow2)
> 
> You can look at how this was done for -cpu
> 
> you do -cpu ?model
> 
> and it list the models.

We do have something like this for block options, -o ? displays this
information - however, it displays them only for a given format, just
like cpu -? displays them only for one architecture.

What is different with my suggestion is that you would iterate over all
image formats and coalesce duplicates into a single line. It shouldn't
be really hard, but a little more work than just changing the help
string to contain some examples.

Kevin