Re: [PATCH 10/21] q800: add easc bool machine class property to switch between ASC and EASC

[Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>]

On 11/09/2023 06:15, Markus Armbruster wrote:

> Philippe Mathieu-Daudé <philmd@linaro.org> writes:
> 
>> On 8/9/23 08:54, Mark Cave-Ayland wrote:
>>> On 07/07/2023 09:29, Philippe Mathieu-Daudé wrote:
>>>
>>>> On 2/7/23 17:48, Mark Cave-Ayland wrote:
>>>>> This determines whether the Apple Sound Chip (ASC) is set to enhanced mode
>>>>> (default) or to original mode. The real Q800 hardware used an EASC chip however
>>>>> a lot of older software only works with the older ASC chip.
>>>>>
>>>>> Adding this as a machine parameter allows QEMU to be used as an developer aid
>>>>> for testing and migrating code from ASC to EASC.
>>>>>
>>>>> Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
>>>>> ---
>>>>>    hw/m68k/q800.c         | 30 +++++++++++++++++++++++++++++-
>>>>>    include/hw/m68k/q800.h |  1 +
>>>>>    2 files changed, 30 insertions(+), 1 deletion(-)
>>>>
>>>>
>>>>> +static bool q800_get_easc(Object *obj, Error **errp)
>>>>> +{
>>>>> +    Q800MachineState *ms = Q800_MACHINE(obj);
>>>>> +
>>>>> +    return ms->easc;
>>>>> +}
>>>>
>>>> Is the getter useful? Otherwise:
>>> Isn't it a requirement? Otherwise I can see that if we decide to enumerate machine properties (similar as to how device properties appear in "info qtree") then it would be impossible to display its value. Certainly at the moment we consider that adding an object property to an underlying struct effectively makes it "public".
>>
>> Just FYI this is not a requirement, per "qom/object.h":
>>
>>    /**
>>     * object_property_add_bool:
>>     * @obj: the object to add a property to
>>     * @name: the name of the property
>>     * @get: the getter or NULL if the property is write-only.
>>     * @set: the setter or NULL if the property is read-only
>>
>> I'm not sure when we want a write-only QOM boolean property, so I
>> genuinely ask, since I agree introspecting QOM object fields from
>> the monitor is helpful.
> 
> I suspect write-only properties came out of QOM's generality curse.  Do
> we have even one?  QOM's design makes this somewhat to tell.

Good question. Given that it's towards the beginning of the next dev cycle, perhaps 
it is worth sending a patch to find out? ;)


ATB,

Mark.