User bus conclusion

[Lennart Poettering]

On Wed, 10.11.10 16:54, Thiago Macieira (thiago at kde.org) wrote:

> Em Quarta-feira, 10 de Novembro de 2010, às 16:47:40, Lennart Poettering 
> escreveu:
> > > I gather from another email that there's a solution for this with systemd
> > > and  loginuid. However, since systemd is not everywhere today nor
> > > required, we must consider the case where it's not present.
> > 
> > Well, auditing has been existing since quite some time it has little to
> > do with systemd. Now, systemd will take all the various things such as
> [snip]
> 
> That's a kernel feature. We cannot rely on it.

Well, and so are BSD sockets. And we do rely on it.

> Maybe so for where we really care about, Linux systems. But D-Bus is used 
> elsewhere...

Well, if people think it's a good idea to limit themselves to POSIX then
it's their own problem.

I want to bring Linux to the next level. And to achieve that I'll try
not to break existing stuff (i.e. we will continue to support --session
for a while as I mentioned a gazillion of times right now), but don't
expect me to care that the new stuff we come up with actually works on
all OSes the same way.

That said, Ryan actually found a good counterpart to XDG_RUNTIME_DIR on
windows, and we could make glib and dbus point to that on Windows. Of
course it will not have the identical semantics since there is no
cleanup and you cannot really place sockets there, but it will allow us
to provide similar semantics for the dbus user bus on windows, too.

If the OS is broken, fix the OS. The only OS that cannot be fixed is
Windows, and there's a clear path how to support that.

Lennart

-- 
Lennart Poettering - Red Hat, Inc.