Re: [Qemu-devel] [PATCH v2 2/6] qapi: support nested structs in OptsVisitor

["Kővágó Zoltán" <dirty.ice.hu@gmail.com>]

2015-06-17 10:41 keltezéssel, Gerd Hoffmann írta:
> On Mi, 2015-06-17 at 09:50 +0200, Markus Armbruster wrote:
>> Copying László because his fingerprints are on OptsVisitor.
>>
>> "Kővágó, Zoltán" <dirty.ice.hu@gmail.com> writes:
>>
>>> The current OptsVisitor flattens the whole structure, if there are same named
>>> fields under different paths (like `in' and `out' in `Audiodev'), the current
>>> visitor can't cope with them (for example setting `frequency=44100' will set the
>>> in's frequency to 44100 and leave out's frequency unspecified).
>>>
>>> This patch fixes it, by the following changes:
>>> 1) Specifying just the field name will apply to all fields that has the
>>>     specified name (this means it would set both in's and out's frequency to
>>>     44100 in the above example).
>>> 2) Optionally user can specify the path in the hierarchy. Names are separated by
>>>     a dot (e.g. `in.frequency', `foo.bar.something', etc). The user need not
>>>     specify the whole path, only the last few components (i.e. `bar.something' is
>>>     equivalent to `foo.bar.something' if only `foo' has a `bar' field). This way
>>>     1) is just a special case of this when only the last component is specified.
>>> 3) In case of an ambiguity (e.g `frequency=44100,in.frequency=8000') the longest
>>>     matching (the most specific) path wins (so in this example, in's frequency
>>>     would become 8000, because `in.frequency' is more specific that `frequency',
>>>     and out's frequency would become 44100, because only `frequency' matches it).
>>
>> Can you explain why the complexity is needed, i.e. why we can't just
>> require full paths always?
>
> Keeping the short names is required for -netdev backward compatibility.
>
> Restricting to short or full (i.e. something= or foo.bar.something=, but
> disallow bar.something=) should not be a problem.  I'm not sure this
> simplifies things much though.  We have to build the full path anyway,
> and I think bar.something= is just a convenient thing we get almost for
> free ...

With the current implementation you can specify (see my previous patch) 
in.try-poll=off in case of alsa. If we would need full paths, it would 
look like opts.data.in.try-poll=off, which is probably not something we 
want.