On 12/07/2015 08:55 PM, Eric Blake wrote:
> visit_start_struct() and visit_type_enum() had a 'kind' argument
> that was usually set to either the stringized version of the
> corresponding qapi type name, or to NULL (although some clients
> didn't even get that right).  But nothing ever used the argument.
> It's even hard to argue that it would be useful in a debugger,
> as a stack backtrace also tells which type is being visited.
> 
> Therefore, drop the 'kind' argument as dead.  While at it, change
> the signature of visit_start_struct() to place the 'name'
> argument at the end (other than 'errp'), and the 'size' argument
> next to 'obj'; this placement of 'name' matches matches how all
> other functions in visit.h do it (visit_type_enum() places
> 'strings' between 'obj' and 'name'; visit_get_next_type() places
> 'promote_int' between 'type' and 'name').  This also avoids the
> confusion caused by splitting related pieces of information,
> where the old signature an unrelated parameter in between the
> "typename" and sizeof(typename) arguments.

I should probably spell it out better in the commit message; I was going
from:

visit_start_struct(v, obj, [kind,] name, size, err)

to:

visit_start_struct(v, obj, size, [kind,] name, err)

then dropping kind as unused.  But I'm seriously thinking about doing
the argument shuffle in the opposite direction (move name earlier,
rather than later):

visit_start_struct(v, name, obj, [kind,] size, err)

with a global coccinelle patch that changes ALL visit_type_* to put name
before obj, because after all we are tying this to JSON which uses
"name":value and it looks odd that every one of our visitor functions
takes parameters in 'value, name' order.

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org