[patch] generate marshallers and metadata from dbus-glib-tool

[Havoc Pennington]

On Tue, 2004-11-09 at 20:59 -0500, Colin Walters wrote:
> To be a bit more precise, the distinction is ownership of the memory,
> not dynamic/static allocation.  One approach would be to just always
> assume the marshallers own the string, but that would kind of suck if
> you were were exporting say a text widget with a GetBuffer method, that
> would normally return a const pointer to a potentially very large
> buffer.

I think we should have the consistent convention that out params have to
be freed by the caller. Just seems simpler as a start. I suppose GObject
introspection framework, whatever it is, will surely support 
"const char* foo_get_name()" type of stuff though.

>   On the other hand - DBusMessage already copies everything you
> stick in to it (something I think we should not do as an option for byte
> [] at least, but that's another issue).

Two options:

 - add _unswapped() variants to the getters, and then either 
   message_get_byte_order() or message_has_native_byte_order()

 - always return by const reference; if you call a getter 
   on a field/arg where byte order matters, and the message is in the 
   wrong byte order, swap the *entire* message in place and then 
   return the const ref. i.e. byteswap lazily.
   There are possible odd side effects of this, since reading would 
   involve writing.

Havoc