dbus-java on windows

Simon McVittie simon.mcvittie at collabora.co.uk
Mon Jan 19 11:13:39 PST 2015


On 19/01/15 18:09, Keresztfalvi, Laszlo wrote:
>   * http://www.matthew.ath.cx/ is no longer available and getting the
>     parts of libmatthew-java for windows is a problem

You are combining a couple of things (D-Bus from Java; D-Bus on Windows)
that have some sort of support, but have never been among D-Bus' "core"
uses; so I'm afraid the best case you can expect is that there are some
rough edges, and the worst case is that nobody has ever tried it.

I have no idea about the Java-specifics or whether they ever worked on
Windows. D-Bus on Unix normally uses AF_UNIX sockets, which are not
available on Windows (D-Bus on Windows has to fall back to TCP instead);
so I would not be surprised if dbus-java needed recompilation with
different options, or internal modifications, to disable the Unix socket
dependency and hence work on Windows.

You might be able to find snapshots of dbus-java's dependencies by
following links from <https://packages.debian.org/unstable/libdbus-java>
(the "orig.tar.gz" link is the source code).

> It'd be great to have dbus-daemon in Java on Windows too but I can run
> it from the successfull native libdbus build.

It is not necessary for dbus-daemon to be written in the same language
as your client process.

The reference dbus-daemon shipped with libdbus is by far the most-tested
implementation. It is also the reference implementation that accompanies
the specification, so it is highly unlikely that any other
implementation has features that are interoperable between languages,
but are not also present in the reference dbus-daemon.

The other dbus-daemon implementations I am aware of are one in GLib and
one in dbus-sharp; they might work for basic functionality, but I would
not recommend them for production use. (The one in GLib is mostly there
so that GLib's regression tests can be self-contained, as far as I
understand it.)

    S



More information about the dbus mailing list