Bug 84289

Summary: Include accessibility bus as one of the well-known message bus
Product: dbus Reporter: Alejandro Piñeiro (freenode IRC: apinheiro) <apinheiro>
Component: coreAssignee: D-Bus Maintainers <dbus>
Status: RESOLVED MOVED QA Contact: D-Bus Maintainers <dbus>
Severity: normal    
Priority: medium CC: mgorse
Version: unspecified   
Hardware: Other   
OS: All   
Whiteboard:
i915 platform: i915 features:

Description Alejandro Piñeiro (freenode IRC: apinheiro) 2014-09-24 16:36:01 UTC
From dbus documentation [1] you can read the following:

"The well-known type of the message bus. Currently known values are "system" and "session";"

But since the dbus based at-spi2 entered to replace the CORBA based at-spi, most of the linux systems are using a custom accessibility message bus. It is explained on at-spi2 README file [2] and on the (outdated) linux foundation wiki.

I think that it is worth to include accessibility as one of the well know message bus. Current situation is somewhat confusing if you start to look to the configuration file and what their options mean (for example, while reviewing [4]) and you find that on type, only session and system are supposed to be allowed.

Unless, at-spi2 is using wront that <type></type> option. If so, any advice is welcome.


[1] http://dbus.freedesktop.org/doc/dbus-daemon.1.html
[2] https://git.gnome.org/browse/at-spi2-core/tree/README#n93
[3] http://www.linuxfoundation.org/collaborate/workgroups/accessibility/atk/at-spi/at-spi_on_d-bus#buslauncher
[4] https://bugzilla.gnome.org/show_bug.cgi?id=722738
Comment 1 Simon McVittie 2014-09-25 13:03:41 UTC
(In reply to comment #0)
> I think that it is worth to include accessibility as one of the well know
> message bus.

I had assumed that the accessibility bus is a private implementation detail of AT-SPI. Is this the case, or is it designed to be an API that can be relied on, or what?

> Unless, at-spi2 is using wront that <type></type> option.

Maybe. See the linked AT-SPI bug.

The options seem to be:

* document the accessibility bus in the D-Bus Specification, making it
  into a third well-known bus

* document somewhere (dbus-daemon man page?) that "accessibility" is
  reserved for AT-SPI

* omit the <type> for AT-SPI
Comment 2 Simon McVittie 2014-09-25 13:12:41 UTC
dbus/doc is a trap for the unwary, referring to a long-dead project to write an IDL or documentation format for D-Bus APIs, and the D-Bus maintainers don't get its bugmail. Reassigning to dbus/core which is the home of everything in dbus.git.

(I asked the fd.o sysadmins to fix this some time ago.)
Comment 3 Simon McVittie 2014-10-28 13:19:04 UTC
(In reply to Simon McVittie from comment #1)
> The options seem to be:
> 
> * document the accessibility bus in the D-Bus Specification, making it
>   into a third well-known bus
> 
> * document somewhere (dbus-daemon man page?) that "accessibility" is
>   reserved for AT-SPI
> 
> * omit the <type> for AT-SPI

So which of these do you want?

If it doesn't break anything, I would personally suggest omitting the <type>, which would be an at-spi patch (in which case this bug can be closed NOTABUG).

However, on https://bugzilla.gnome.org/show_bug.cgi?id=722738 Magdalen Berns wrote "I went ahead and tested removing <type> but that seemed to break
guis, so it's a no go". I'd be curious to hear what breakage it caused and why. On the GNOME bug I wrote:

> The only practical effects of the <type> at the moment are:
>
> * activated services get DBUS_STARTER_BUS_TYPE=accessibility in their
>   environment
>
> * if the type is exactly "system" or "session", it shows up in syslog
>   message prefixes

but perhaps I missed some other practical effect?

If there is a reason to prefer one of the first two options, please describe that reason in the commit message of a suitable patch for dbus :-)
Comment 4 Magdalen Berns 2014-11-01 15:49:08 UTC
(In reply to Simon McVittie from comment #3)
> (In reply to Simon McVittie from comment #1)
> > The options seem to be:
> > 
> > * document the accessibility bus in the D-Bus Specification, making it
> >   into a third well-known bus
> > 
> > * document somewhere (dbus-daemon man page?) that "accessibility" is
> >   reserved for AT-SPI
> > 
> > * omit the <type> for AT-SPI
> 
> So which of these do you want?
> 
> If it doesn't break anything, I would personally suggest omitting the
> <type>, which would be an at-spi patch (in which case this bug can be closed
> NOTABUG).
> 
> However, on https://bugzilla.gnome.org/show_bug.cgi?id=722738 Magdalen Berns
> wrote "I went ahead and tested removing <type> but that seemed to break
> guis, so it's a no go". I'd be curious to hear what breakage it caused and
> why. On the GNOME bug I wrote:
> 
> > The only practical effects of the <type> at the moment are:
> >
> > * activated services get DBUS_STARTER_BUS_TYPE=accessibility in their
> >   environment
> >
> > * if the type is exactly "system" or "session", it shows up in syslog
> >   message prefixes
> 
> but perhaps I missed some other practical effect?
> 
> If there is a reason to prefer one of the first two options, please describe
> that reason in the commit message of a suitable patch for dbus :-)

I noticed that I couldn't open graphical applications. However I am not certain I checked this thoroughly enough to say with absolute certainty this was the result of removing type. To be sure, I will have another look and confirm the state of things just after doing that.
Comment 5 GitLab Migration User 2018-10-12 21:21:25 UTC
-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/dbus/dbus/issues/113.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.