Re: [Qemu-devel] [PATCH 4/8] qdev/core: bus list

[Markus Armbruster <armbru@redhat.com>]

Anthony Liguori <anthony@codemonkey.ws> writes:

> Markus Armbruster wrote:
>> Paul Brook <paul@codesourcery.com> writes:
>>
>>   
>>> On Tuesday 30 June 2009, Markus Armbruster wrote:
>>>     
>>>> Paul Brook <paul@codesourcery.com> writes:
>>>>       
>>>>> On Tuesday 30 June 2009, Gerd Hoffmann wrote:
>>>>>         
>>>>>>  * maintain a list of busses.
>>>>>>  * maintain bus numbers.
>>>>>>  * add function to find busses by type / name / number.
>>>>>>  * add monitor command to list busses.
>>>>>>           
>>>>> I still object to this patch. Busses should be identified by their
>>>>> location in the tree, not by number.
>>>>>
>>>>> Paul
>>>>>         
>>>> Location in the tree can be uniquely identified by a number.  Handy when
>>>> all you want is enumerate the buses, and you don't really care where
>>>> they're hanging out in the tree.  Why should something like that not be
>>>> done?
>>>>       
>>> The address of the BusState is also a locally unique
>>> identifier. That doesn't mean it's a good thing to expose to the
>>> user.
>>>
>>> Paul
>>>     
>>
>> Red herring.
>>   
>
> I don't think that's a very useful response.
>
> I think it's a perfectly valid suggestion that we should identify
> buses based on the their location in the tree to users verses a number
> generated based on some hashing algorithm.
>
> A tree location has meaning to a user.  A random integer doesn't.

Numbering nodes according to a well-defined tree traversal is not
random.  We can discuss whether using such a number in an interface is a
good idea (nobody suggested to use it *instead* of tree paths).
Rejecting the idea by comparing it to some internal address, however, is
not a very useful contribution to such a discussion.

But I figure I'm wasting my time and yours, so I'll shut up now.