On Tue, Jan 4, 2011 at 3:57 PM, Marcel Holtmann <[email protected]> wrote:
> Hi Denis,
>
>> >>
>> >> dbus_message_iter_append_basic(&entry, DBUS_TYPE_STRING,
>> >> &(val_array[i + 0]));
>> >> - dbus_message_iter_append_basic(&entry, type,
>> >> - &(val_array[i + 1]));
>> >> + if (type == DBUS_TYPE_STRING) {
>> >> + dbus_message_iter_append_basic(&entry, type,
>> >> + &(val_array[i + 1]));
>> >> + } else {
>> >> + dbus_message_iter_append_basic(&entry, type,
>> >> + val_array[i + 1]);
>> >> + }
>> >
>> > this does looks a bit hackish. Can you put it into context please. It
>> > might be that our call chain is wrong.
>> >
>>
>> We only ever needed (and thus ever added) support for string:string
>> dicts. E.g. ones that could be g_strfreev-ed. No other dicts were
>> supported by this function.
>
> so it is the g_strfreev part that makes this hackish. Fine by me then,
> but we might need to comment on why string arrays in this case are
> special.
I don't think so. You use g_strfreev if you are using { string, string
} dict to free all strings. Why is this hackish?
Even if we had another interface that didn't required the g_strfreev,
the issue with calling dbus_message_iter_append_basic() would not
change once we continue using a void** to store pointers to different
types at the same time..
Lucas De Marchi
_______________________________________________
ofono mailing list
[email protected]
http://lists.ofono.org/listinfo/ofono
