Wednesday, March 2, 2011

Nokia says "Meego with another OS" and Terminal Mode goes "Terminal"

Ok, how about a little controversy.  I won't bother spending too much time discussing Nokia's choice to drop Meego and that choice's impact on GenIVI.  In fact, I will say that it is just one more blow to the organization, that in my opinion lost any hope of broad market adoption about a year ago, but in any case, they will just pick up and move on to the next GUI / Application framework.  There was already debate within GenIVI if Meego was ever their "official" direction to begin with.  Likely, they might pick something a little more credible in automotive.  Frankly, even though QT is also associated with Nokia, I think it is more likely that QT will continue to have a life, even given Nokia's selection of Microsoft as their phone platform.  I wouldn't be surprised to have QT attempt to resuscitate life into GenIVI working groups, but still don't hold out much hope.

I think what I am more interested to share are my views on the impact of the Nokia decision to Terminal Mode.  Unlike GenIVI, Terminal Mode was gaining a bit of traction, or more specifically, not Terminal Mode per se, but VNC.  There were some very fundamental problems with Terminal Mode, for example since Terminal Mode lacked a great policy management story (what is allowed to be displayed when a vehicle was in motion) every tier 1 was implementing their own policy management solution.  This created issues for OEMs looking to implement solutions across multiple Tier 1s or multi-sourced production programs. 

Not being intimately involved in the working group, I will say my limited understanding was that the Terminal Mode supporters were looking to address this issue by creating a consortium of Terminal Mode participants that could address policy management and rank applications on a scale of "appropriateness" to use while the vehicle was in motion.  Since this appeared to be based on self certification by application vendors, this approach seemed questionable, to me, for automotive use.

Another major challenge to Terminal Mode was performance.  I think this is where the choice by Nokia has the most significant impact to Terminal Mode.  For anyone who has seen a Windows Phone 7 user interface, you would know that UI is in constant motion and transition.  It is the same UI that in a recent survey by Nokia is "one of the most exciting aspects of Microsoft deal" according to a Nokia poll released this week. Click here to see the Engadget Article

Already Terminal Mode was being bashed by their own supporters of the VNC approach, even within the CE4A.  The reason, performance, but such an animated, constantly moving UI like the one in Windows Phone 7 would, in my opinion, grind already sketchy performance to a halt. 

I have no doubt that there might be some creative engineers at Nokia who could have addressed the performance issues of Terminal Mode in a Windows Phone 7 environment, but I have a feeling Microsoft might have ideas on a more elegant application / screen sharing solution.  So without Nokia backing Terminal Mode, I think you see a short term VNC approach as an interim solution for sharing specific applications from a phone to a head unit, but I don't see formal Terminal Mode existing in the long term.

So what do I think will be the right solution?  Video (like iPod Out) or HTML (like Blackberry Bridge).

I will do a more in depth blog about iPod Out vs Terminal Mode vs HTML / Blackberry Bridge vehicle integration soon.

Thanks to Alistair and Jorg from Nokia let me provide a few links to Terminal Mode if you want to learn more.  I appreciate the education guys:

To learn about the Terminal Mode event March 23rd in Japan:
To learn more about Terminal Mode in general:


  1. In your first paragraph you are confusing MeeGo with application frameworks. MeeGo is a Linux distribution that happens to use Qt as the application framework. Qt also runs on any other Linux distribution as well as your very own QNX and Windows Auto.

    This distinction is important as GENIVI does not specify a GUI toolkit so your comment "they will just pick up and move on to the next GUI / Application framework" holds no water.

    You did though get one thing right in the first paragraph, Qt will continue to have a very vibrant life.

  2. Hey Alistair good point. I was under the impression that it was originally specifying / using a different ui / app kit like Moblin or Mentors. What do you think of terminal mode? You made some good points in your email about animations with Silverlight. Feel free to post them here too.

  3. Given that Android is now the #1 global smartphone platform, and only likely to increase its penetration, the smart choice for automotive manufacturers would be to follow Saab's lead and adopt Android.

    This would have the benefits:
    1. Familiar UI
    2. Fast track time to market
    3. Open source licenses that prevent lock-in

    The alternative is to reinvent the wheel, and that is not a smart strategy.

  4. OK, you asked ;-)

    You said in your email "Silverlight over vnc would be awful performance".

    I replied "It's not Silverlight over VNC. VNC doesn't care about care about the application framework, it's just transferring pixels."

    I inferred that by referring to Silverlight you were perhaps meaning the animated nature of the WP UI. A phone experience is intended to be an immersive experience which is the last thing you'd want in front of a driver. All the guidelines from around the world for driver UI recommend against animations.

  5. Hi, I am Jörg from Nokia. I am have been leading the specification work of Terminal Mode in the collaboration with the CE4A. Interesting reading, but I want to comment on some parts which need some clarifications/corrections.

    Based on recent discussion, posts and conference, I disagree that it is VNC rather than Terminal Mode that is gaining traction. I see many companies actively exploring, prototyping and developing Terminal Mode solutions and I had the pleasure be on a panel with you at the latest QNX conference in Stuttgart, speaking about Terminal Mode opportunities. Hasn’t even QNX making a good Terminal Mode support video Anyway, I believe we do agree that VNC as such is not sufficient alone in an automotive setup to ensure driver distraction-free usage of mobile device applications. Therefore we have put additional mechanisms and elements into Terminal Mode to facilitate this. Self-certification from application vendors is not going to be part of this. The details of these mechanisms can be found by downloading the freely available specifications from

    With regard to user interfaces: Please have in mind that Terminal Mode enables the transfer of the application’s user interface into the in-vehicle infotainment system. I do not see that automotive applications will have the same flashy-style user interface, as regular applications. Rather, I expect that automotive applications automatically switch to a driver-friendly user experience when detecting that they are being used while driving. This is an opportunity for application developers. Terminal Mode does provide this functionality.

    One final comment on Nokia’s support for Terminal Mode. Our recent announcements have not changed any of our commitment and plans for Terminal Mode. I hope you will be convinced when seeing our first Terminal Mode enabled products later the year.

    Looking forward to more discussions on this exciting topic and why not come to the first Terminal Mode summit of the Terminal Mode consortium in Tokyo on March 23 to learn more about the specification in detail and meet all the industry participants in the Terminal Mode consortium. Details at

  6. Hi Alistair and Jorg!

    Thank you so much for your comments and calls. I really appreciate the education and enthusiasm.

    Alistair, I know a phone UI should be different from what is exported to the vehicle, but often you have OEMs, Analysts, etc. that think of Terminal Mode or other screen sharing mechanisms as a silver bullet to avoid porting costs to get many different applications in a vehicle. Once you start tailoring the apps in the phone (either for distraction or performance reasons) some of the value is lost. I recently spoke to an analyst that was talking about the cost of porting an application to automotive being approximately $100k / app and his comment was terminal mode would remove this cost. I don't think so. In many cases this is similar to the debate we had with GenIVI.

    GenIVI was established with many OEMs frustrated that the porting costs for VR, TTS, etc. in NRE was incurred again and again with every new program launched. I believe they incorrectly associated that costs to porting to "proprietary OSes" when often that cost was effectively a qualification of a "Gold Master" solution that took into consideration of the specific OS version, Audio Path, Hardware Platform, etc. We have heard some crazy costs associated with application porting associated with GenIVI.

    Also, I know I intermix VNC and Terminal Mode in a few places. To be very clear in every production program I am tracking the Tier 1 or OEM seems to be using RealVNC on the head unit. That way they can support a Terminal Mode enabled phone or load a VNC server onto a Blackberry or Android device and also support non-Terminal Mode devices.

    That is why I say it has been more VNC success on the head unit side. That being said Terminal Mode is what drove the initial interest.

    The Android comment I will have to come back to. I am not convinced Android is an appropriate OS for automotive use. If Google were to support automotive directly, I might have a different reaction, but I can post more soon about that.

    I will link to the terminal mode events in my original post. Thanks again!

  7. P.S. Jorg, you mentioned on the phone some thoughts around a higher level UI. Care to comment here about what that might be? Please say HTML. =)

  8. May be best to start with why we have been using VNC in the first place.

    It does offer a couple of benefits, I want to highlight (besides others):
    - Simple protocol
    - Allows for implementation even on mid to lower end head-units
    - No dependency to the Operating System, neither on the mobile device, nor on the head-unit
    - No dependency to the application framework, as it uses the framebuffer (i.e. pixel data) as an abstraction layer

    Some personal comments on the higher layer of abstration: Of course I do see discussions around other ways to abstract a user interface. As you can guess there any many possibilites, HTML being one of them... The disadvantage I do see in those cases is that it creates some dependencies on operating systems, application frameworks or run-times used. So, even if you would define a higher-level abstraction layer, you will depend on a fallback mechanism, like VNC.

    I believe that we do have all the elements together in Terminal Mode, to enable all the relevant use cases. We have to have in mind, that while driving we are not watching a video, we are not playing a game and we should not expect a flashy user interface, but one, which brings the relevant information to our attention with as less distraction (e.g. glance time) as possible.

    Hei, I hope I encouraged you enough to join the consortium and actively particiate the discussions there :-).

  9. Hi, probably our entry may be off topic but anyways, I have been surfing around your blog and it looks very professional.

    It’s obvious you know your topic and you appear fervent about it. I’m developing a fresh blog plus I’m struggling to make it

    look good, as well as offer the best quality content. I have learned much at your web site and also I anticipate alot more

    articles and will be coming back soon.

    Infotainment Systems