Replacing evil looping code

Bastien Nocera hadess at hadess.net
Mon Nov 6 07:47:55 PST 2006


Hey Havoc,

On Mon, 2006-11-06 at 10:12 -0500, Havoc Pennington wrote:
> Hi,
> 
> > Is it possible to replace all that crap with decent D-Bus code?
> > 
> > I don't think we can use a service, as we're depending on some
> > command-line arguments (should we pass those to the service in some
> > other way instead?).
> 
> Remember you can autostart a service by just sending it a method call, 
> so if your command line args can be args to a method that might work.

I figured that much. But see below.

> The showstopper for using a service in your case looks like the dynamic 
> bus name creation (putting the pid in there), there's no way to put the 
> dynamic bus name in the .service file.

To be fair, we don't really need that "PID" bit, but we'd need one
instance of the service per instance of our plugin in Mozilla (well,
pretty much).

If we can that one-to-one relationship without using a dynamic service
name, it wouldn't bother me one bit to use it. I didn't see any ways to
have that using the service route.

> If you figured a way around that, then just calling 
> dbus_connection_send_with_reply_and_block() (or glib equivalent) would 
> be what you want; it would send a method call, which would block until 
> the service started up, and then return.

Nod.

> Otherwise, your code doesn't seem that horrible to me, it's pretty much 
> the usual "block in main loop code" (looks about the same as 
> gtk_dialog_run() for example). I usually try to avoid this and just do 
> things asynchronously, but I'm guessing your code is in some scary 
> mozilla context that makes that difficult...

Well, Christian Persch really dislikes it, and I can fully understand
why :)

> dbus could have a dbus_connection_block_for_signal() call, but it might 
> be somewhat dangerous; unlike method call replies, signals aren't 
> usually very reliably known to be about to arrive, and dbus does not 
> block in a reentrant way so the app would be locked up...

Expecting signals is always a bit worrying, when you're not 100% sure
it's going to arrive :)

As I said, if there was a way to differentiate instances of the same
service, we'd be in business using the services activation.

Any ideas welcome.

Cheers



More information about the dbus mailing list