[OTR-dev] Using libotr

[Martin Milata]

On Tue, Mar 11, 2014 at 23:33:37 +0000, Alex Burzyński wrote:
> Hi Martin,
> 
> Thanks for answering, but that triggers more questions!
> 
> I'm trying to draw some use-cases for those different policies and decide
> if they even should be exposed to the end-user.

I can't think of a situation where you'd want to expose them to a chat
application user. I'd even say that you should use
OTRL_FRAGMENT_SEND_ALL unless you have a reason not to.

> So as I understand the fragmentation, the plain text is encrypted and then
> split into fragments.
> 
> And in my mind the app's inject_message callback merely sends the message
> fragments to the other end, there shouldn't be any extra logic in there.
> convert_msg looks like a place for that extra logic.
> 
> So in case of ALL_BUT_LAST what else could the app do with the last
> fragment other then send it? I'd imagine that without the last fragment the
> message wouldn't be readable at all on Bob's end.
> 
> Could you explain what that hook you've mentioned might want to do with
> that last fragment?

What I meant is that ALL_BUT_LAST is useful when writing plugin for
existing application. Common pattern is that the plugin registers a
callback/hook that the application calls whenever a message is sent:

string on_message_send(string message_to_be_sent) {
    ...
    otrl_message_sending(..., &lastfragment, OTRL_FRAGMENT_SEND_ALL, ...);
    ...
    return lastfragment; // application expects the callback to return
                         // modified message
}

Here's a real-world example where this might be useful:
http://lists.cypherpunks.ca/pipermail/otr-users/2007-September/001124.html

Hope that makes some sense:)
Martin

> Thanks,
> Alex
> 
> On 11 March 2014 12:06, Martin Milata <b42-ml at srck.net> wrote:
> 
> > Hi,
> > not sure about the other questions but perhaps I can at least answer
> > this one:
> >
> > On Tue, Mar 11, 2014 at 01:14:53 +0000, Alex Burzyński wrote:
> > > Fragmentation policy
> > > I don't fully understand the how
> > > OTRL_FRAGMENT_SEND_ALL_(BUT_FIRST|BUT_LAST) work, where those fragments
> > are
> > > stored and why?
> > > how my use of otrl_message_sending/receiving() should change depending on
> > > OtrlFragmentPolicy?
> >
> > When you use one of the two policies, the fragment that is not sent is
> > returned in the messagep pointer passed to otrl_message_sending. The
> > other fragments are sent using the inject_message callback you provided.
> >
> > The OTRL_FRAGMENT_SEND_ALL_BUT_LAST is useful when you are calling
> > otrl_message_sending from some kind of a hook that expects you'll have
> > modified message to send at the end (i.e. the application itself handles
> > the last fragment). Not sure when OTRL_FRAGMENT_SEND_ALL_BUT_FIRST can
> > be useful.
> >
> > Hope that helps,
> > Martin