Starting discussion on a new version of the notification spec

Christian Hammond chipx86 at
Sat Jun 13 15:08:34 PDT 2009

2009/6/13 Aurélien Gâteau <aurelien.gateau at>

> Brian J. Tarricone wrote:
> > On 06/13/2009 02:18 PM, Aurélien Gâteau wrote:
> >> Brian J. Tarricone wrote:
> >>> 1.  Passive vs. active notifications.  I recall that notify-osd
> >>> unilaterally decided that the 'actions' bit of the spec was Bad[tm] and
> >>> that notifications should be entirely passive and not accept input.
> >> I would rather not start a discussion on this subject: it has been
> >> debated to death and people won't change their mind.
> >
> > That's rather closed-minded.  But I suppose if Canonical wants to go
> > their own way and ignore community consensus, it's free to do so.
> What I wanted to say is that I read a lot of discussions on that issue
> and I learned nothing positive ever come out of those, so I'd rather
> talk about icons, markup and other things instead.
> The spec says actions are optional and I don't think Canonical wants to
> have this changed.
> (Also note that I have personally nothing to do with the choice to go
> action-less, I was not even working for Canonical at the time of this
> decision)
A little bit of background on this.

The intent originally was to make it optional so that a daemon could
intercept the messages and log to console or to a log file, which wouldn't
allow for feedback. The original notification-daemon was actually built this
way, to be able to be used in a graphical or console manner.

We intended for all graphical implementations to support actions. The spec
does not explicitly say this, which is an oversight. At the time there was
just libnotify and notification-daemon, so it didn't really matter.
notification-daemon was considered the reference implementation that
provided the minimum amount of functionality a daemon should implement.

If I thought we'd face an issue down the road where actions would be removed
from all apps in a distro and the new notification-daemon replacement would
not support them, I would have made this *VERY* clear in the spec from the
beginning. I think it's harmful to do this.


Christian Hammond - chipx86 at
Review Board -
VMware, Inc. -
-------------- next part --------------
An HTML attachment was scrubbed...

More information about the xdg mailing list