Null Value Type?

Havoc Pennington hp at redhat.com
Mon Aug 6 07:58:08 PDT 2007


Hi,

jamie wrote:
> this would not break anything as it would be a new data type - is this
> feasible to have in a future dbus version?
> 

See the notes I posted to the list from just after GUADEC. My favorite 
suggestion (though I would expect to learn more from further discussion 
and prototyping) was the maybe-types approach that Ryan proposed, like 
"ms" = "maybe-string" i.e. string-or-null

There is significant work though, both to add feature negotiation as 
Simon mentioned and then to add the new types to libdbus and the various 
bindings. Both writing up and following through on a discussion of the 
proposed changes to the spec, and then writing code.

This is basically true of all the stuff mentioned in those GUADEC notes, 
that we need volunteers.

As I said last week, to the extent that I have extra time, I consider 
the most pressing issues to be one level up, having enough convenience 
API and fixing GNOME's usage of dbus. That's where I see the most lines 
of code in apps and the most bugs in apps (doing name ownership and 
introspection and so forth incorrectly).

Right now it takes most of my dbus time allotment just to try and avoid 
being the bottleneck for other people's patches though.

In short, you should not count on any particular timeframe for this 
feature or any other unless you code it yourself and then are very 
persistent about getting the patch reviewed ;-)

Havoc



More information about the dbus mailing list