Todo items that need to be looked at for 0.90

[Havoc Pennington]

John (J5) Palmieri wrote:
> Here is the short list of todos that need to be looked into before we
> can do a 0.90 release, with comments:
> 
> 
> Group DBusAuth 
>         some SASL profiles require sending the empty string as a
>         challenge/response, but we don't currently allow that in our
>         protocol.
>         
> Does our protocol still not allow this.  What would it take to make this
> work and is it a blocker for 0.90?
> 

IIRC the issue is that the format is just
KEYWORD ARG

i.e. not quotes or anything, so empty string looks like you just forgot 
the arg:
KEYWORD

The protocol could be changed to e.g. require quotes on that arg. I 
think there's probably a "quoted string parser" already (used for e.g. 
signal match rules? I don't know) if so that could simply be used.

Since there's no current sasl profile we rely on that requires this, I 
think it could be punted if necessary and some ugly protocol versioning 
or special keyword used to fix it later. But it's perhaps simple to fix 
so could be worth doing.

> ====================================
> 
> Global dbus_set_error 
>         should be called dbus_error_set()
> 
> Global dbus_set_error_const 
>         should be called dbus_error_set_const()
> 
> Ouch.  These are going to cause massive pain (should have caught this
> sooner).  Should we deprecate or leave as is?

Eh, may as well just leave it.

> ====================================
> 
> Global dbus_message_append_args 
>         support DBUS_TYPE_STRUCT and DBUS_TYPE_VARIANT and complex
>         arrays
> 
> Global dbus_message_get_args 
>         support DBUS_TYPE_STRUCT and DBUS_TYPE_VARIANT and complex
>         arrays
> 
> This can be added later but in reality it is a convenience function and
> we can live without bloating the core.
>         

I think these are OK as long as we have a good warning if you try to use 
them on those types. I imagine we do.

Havoc