[Patch] Moving FD passing into the header

[Thiago Macieira]

Em Sexta-feira 18. Junho 2010, às 06.08.52, Marcel Holtmann escreveu:
> > > QDBusPendingReply<ReturnType> reply = interface.asyncCall("something");
> > 
> > You can bind it as reply.wait();  reply.getUnixFileDescriptor(0); right?
> 
> so within BlueZ, obexd, oFono and PulseAudio we were planning to
> actually make a lot of use of the fd passing. So I would say it is very
> much common. And why not?
> 
> As I said before, I really like having the file descriptors as part of
> the introspection data. Treating it as side channel or out-of-band data
> seems pretty much wrong to me. It leads to hacks and other weird per
> application agreements. While once it is part of the actual API, it can
> be properly specified with common D-Bus introspection for methods.

Here's another reason:

what happens if the caller doesn't remember to send the file descriptor in a 
call?

With the signature: the call won't be delivered, resulting in probably an 
UnknownMethod error.

Without the signature: the call will be delivered, and it's up to the callee 
to find out that the file descriptors were not passed, then send its own return 
error.

Also note that removing the file descriptor from the signature makes it 
possible to call a file-descriptor-receiving method from a transport that 
doesn't support passing file descriptors.

Finally, with the addition of Maybe types, we are changing the type system 
anyway.

-- 
Thiago Macieira - thiago (AT) macieira.info - thiago (AT) kde.org
  Senior Product Manager - Nokia, Qt Development Frameworks
      PGP/GPG: 0x6EF45358; fingerprint:
      E067 918B B660 DBD1 105C  966C 33F5 F005 6EF4 5358
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 190 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.freedesktop.org/archives/dbus/attachments/20100618/c1fbde46/attachment.pgp>