Re: [Qemu-devel] [PATCH v6 07/12] qapi: Detect collisions in C member names

[Markus Armbruster <armbru@redhat.com>]

Eric Blake <eblake@redhat.com> writes:

> On 10/02/2015 11:11 AM, Markus Armbruster wrote:
>
>>>> Why wrap function c_name() in a method?  Why not simply call the
>>>> function?
>>>
>>> 'self.c_name()' is shorter than 'c_name(self.name)'.  And I already had
>>> long lines with that seen[self.c_name()].describe() pattern.
>> 
>> You could also try a local variable: cnam = c_name(self.name).
>> 
>>>> It's method in QAPISchemaEntity only because this lets us add special
>>>> cases in a neat way.
>>>
>>> True, but I _did_ mention in the commit message that I did it for less
>>> typing.
>>>
>>> But as to special cases, yes, I have one in mind (although I have not
>>> played with it yet).
>
>> 
>> I'm afraid I don't have enough context to grok this late on Friday :)
>
> Here's another case I have in mind. Right now, we have special code
> littered in qapi-types and qapi-visit to track that the
> QAPISchemaObjectTypeMember for simple unions is named 'type' in QMP but
> 'kind' in C code.  Having member.c_name() return 'kind' would simplify
> that code.

Yes, but naming it 'type' in C will simplify things even more :)

Might make sense as an intermediate step all the same.  However, if this
is the only reason for having a c_name() member, the member will become
superfluous when we clean up the type vs. kind stupidity.

> So, what I will do for v7 is rework the patches to independently
> implement and use member.c_name() (possibly by also creating a special
> subclass of QAPISchemaObjectTypeMember), showing how it makes the
> clients easier, and then this patch can also make use of that work.