Everything and the Mobile Software Universe…

There were rumors (here and here for example) about Apple using an an Intel Atom for its next generation iPhone, well they seems to be wrong: According to this New York Time article who found this out …thanks to a linked-In profile! :

Wei-han Lien, the senior manager of Apple’s chip team, dished out the morsel on LinkedIn, saying he’s busy at work crafting an ARM processor for the next-generation iPhone.

PA Semi NYT article (relayed from Electronista and TUAW)

So Apple, thanks to the 300M$ PA Semi buyout is entering the application processor design, but why? :

For a quick recap the hardware architecture of a smartphone is now pretty standard :

• a "big" cpu sporting all the applications and the high level OS: the application processor
• a 3G/2G modem chipset connected to the application processor, on which the 3G/2G protocol stack is running.
• a mix signal chip for power management and analog to digital conversions
• depending on the modem, a RF chip (connected to the antenna )
• …many various chips (wifi, Bluetooth being the most prominent)

The answer may be as simple as COST: After the phenomenal amount of money spent on 3G chipset modem development the real hardware margin are no more on wireless chipsets, but on the application processor themselves: see the very successful Texas Instrument OMAP line, the Samsung Arm line and others, versus the low margin, high risk 3G/2G chipset business of Infineon for example, or EMP, TI, etc…Qualcomm is a different story with their enormous IPR revenues, around 5% (!) of a mobile phone cost.

By making its own application processor, Apple will retain those margins for itself, while buying modem from other sources…and looking at the iPhone+iPod+whatevernewmobiledevice volumes we are talking here of perhaps hundred of millions of devices a year!  SO the savings, compared to the 300M$ buyout will be quickly amortized.

Of course the other big bonus when doing its own processor is that you can cram in it all the hardware acceleration you need for ….your software, and when your processor will be associated with only ONE OS, the possibility for optimizations (power consumption and speed)  are way above what a generic purpose processor may offer.

Any comments?

Thomas

The Linux community has come up with a bunch of new technologies to bring the UI eye candy on Linux to the next level…following the path of a well established player : Apple with its 8 years old compositing window manager Quartz.

But as it may happen way too often in the Linux desktop community the emphasis is on the technical aspects , not at all  on the ease of use and usability improvements: check the following nice/fancy/eye blowing demos, and after the wow effect … think about what it brings.

Here is an interesting Beryl vs Vista videos: the focus is on effects only, except form the clear Apple borrowing (dock, window in/out effects) the Beryl side is just as unusable as the Vista one can be…but you have eye candy , and a wiz/bang  teenager environment (jackytouch for the frenchies )

Another one with Compiz-Fusion, same conclusion:

…just try to find a utility, a user experience meaning to this fireworks….well nothing, in the same vain as vista windows effects and switcher you deactivate to speed up your PC. Ok the 3D cube is lovely, but come on, playing a 3D game just to switch between workspaces can be called a waste of time, I want to be efficient to do the things I really care on my PC, which ARE NOT looking at the animation between windows!

Then come this little refreshing Linux gem, and, I want to say of course, it is not coming from the "raw" desktop engineering Linux community, but from a lab (ok, a french one ), Inria. It is called Metisse  : and at last it clearly states "Usability study and experiment before nice effects!"

Metisse [1, 2, 3] is an X-based window system designed with two goals in mind. First, it should make it easy for HCI researchers to design and implement innovative window management techniques Second, it should conform to existing standards and be robust and efficient enough to be used on a daily basis, making it a suitable platform for the evaluation of the proposed techniques.

Metisse is not focused on a particular kind of interaction (e.g. 3D) and should not be seen as a new desktop proposal. It is rather a tool for creating new types of desktop environments.

Quote from the Metisse  web site.

Checkout this simple idea about copy/paste: why on earth it is not like that today….

This one shows also some interesting usability concepts:

[SonyEricsson is promoting a new Application Environment mixing Java ME and Adobe Flash Lite: Capuchin. Blogger Thomas Menguy tries to describe it and evaluate what "yet a new" development platform means to the industry ].

Sony Ericsson had a nice webinar last Thursday, interesting held through Adobe E-Seminar:

“Flash Lite meets Java ME on Sony Ericsson phones with Project Capuchin”.

At least now we have some information about Capuchin, and I’ll sum it up for our beloved busy executives:

• A technology that allows developers to make the UI using Flash Lite and code the business logic and access to the platform services with Java (ME).
• A development environment with PC based tools (Adobe CS plugin for flash and Eclipse plugin for Java), simulators and a specific runtime embedded in SEMC phones.
• The deployment is done using the well in place Java deployment environment (jar are used, same signature, etc).

Here is first a transcript of the capuchin webcast, then as a conclusion I’ll throw out my thoughts about this and its impact on the industry (if you are still there…).

Project Capuchin Web Cast transcript

Capuchin is about mixing those two worlds, and enforce UI designers and developers relationship.

Why the Capuchin name : it is a monkey like tamarin…the name of the Adobe Action Script VM.

Here is a high level architecture presentation of Capuchin:

Flash content is embedded into a .jar and can be launched by some Java code, then, thanks to the Capuchin API the Flash Action Script can access the various JSR or any other Java class of the project.

Here is below how an accelerometer API may be available in the Flash Action Script of a Capuchin Application:

The Capuchin API works both way: flash to java and java to flash.

What Capuchin is bringing:

Flash development:

• Extend current limited APIs with the use of JSR
• Secure Flash application
• Deploy flash as java games, distribute Flash content through existing java distribution infrastructures

Java Development

• Clear separation between business code and UI
• Nice development tools
• Professional UI tools

How to use Capuchin:

3 main ways to do:

1. Packaging pure Flash Lite content using jar
2. Java Midlet using Flash Lite for the UI layer
3. Java Midlet using Flash Lite for PARTS OF THE UI

Adobe has a nice technology: mxp, format for packaging extensions. Capuchin use mxp to package the APIs that will be mapped into the Action Script.

There is an Eclipse Capuchin Plugin to create those APIs declaration (see above) as they will be usable in the Action Script written in CS3. This tool outputs an XML file which will be used to output Java Classes for the java part to be implemented …. and Action Script classes to be used in CS3.

Everything is then packaged in a .mxp installation package. SEMC will provide some mxp already (Bluetooth , others…)

Demo time:

The webcast then featured a demo:

swf2jar :

Goal here was to show the tool to convert a swf to a jar, swf2jar: very useful for packaging because a flash game today…end up in the image folder when deployed in a SEMC phone ….

Calendar component:

Project Capuchin plugin for CS3, with mxp packages. The intent here was to show how to use Java services in a Flash Lite content

There are some “Platform components” in the library:  in the AS editor, it is possible to import for example the package com.sonyericsson.capuchin.calendar.Calendar

… to import the Platform classes,  so it is now possible to use the Calendar class as a normal Action Script, even if it is a Java Service.

One word about the toolchain future:

In gray: Not done today:

• Flash Emulator will be connected to Eclipse to use java services directly and not only stubs
• UI library: a flash widget library will be developed
• Connect everything to the existing SEMC phone emulator
• Work with adobe so that in device central, when a SEMC phone is selected the list of available mxp would be provided.

What will be published soon:

• First phone : C905, compatible with capuchin APIs
• Capuchin APIs,Java Classes
• swf2jar tool
• Capuchin API generator , eclipse plugin
• mxp packages with source code
• capuchin test and video tutorials
• demos applications

=>check here http://developer.sonyericsson.com

(final: October)

SEMC Capuchin will be present at Adobe MAX in San Francisco and in Italy in December!

Some Q&A with no major questions…mine were not answered:

• What is the implication of Adobe in this project?
• What is the implication of Esmertec in this project?
• Is there a roadmap to have Capuchin on other platform than SEMC ones?

Some points about this initiative:

• SEMC has already done a large part of their applications in their feature phones in Java, and they have a strong Java commitment with Esmertec, so on SEMC phones Java is the preferred development method internally….and now with Capuchin, externally as nearly all the platform services are already available in Java.
• With the point above, and knowing that some part of the SEMC feature phones themes are already in Flash, merging Flash Lite and Java was a natural choice for SEMC

• Flash Lite choice is the only one possible for today mobile phones (CPU/Memory)…but is really not a complete and efficient UI application frameworks, it lacks …widgets! So SEMC plan to develop some new ones, hum wait, Adobe Flex is not about that? Bringing application development to Flash?

• Not sure about the porting of such a technology on other platforms than SEMC … but from my knowledge only another one has made the Java choice: Google Android where all the platform services can be accessed through Java, but I don’t see any incentive for SEMC to port it to Android

Conclusion

So we have a new Application Environment, with its own SDK, that will certainly be only available on SEMC platforms….Capuchin one will complete this never ending list:

Are we still talking about cross platform development? About consolidation and standardization?  NO

The industry is pushing the other way, and really this is NOT AN ISSUE.

Services and applications developers have learnt how to reuse code across platforms, how to architect their code and services so that it is easy to change only the presentation and the adaptation to the platform: after all developing a UI for a 800*480 screen and a 176*220 is just something completely different, and really not a big deal if your UI is uncorrelated from your services; Capuchin helps that, as many other technologies.

All those new Application Environments are bringing to Mobile Platforms  great core value for services deployment :

• openness
• great tools, ease of development
• focus on user experience and UI
• deployment/packaging/distribution strategies
• security

And we don’t want a “one size fits all” environment, it is simply not true in an industry where the forms factors, capabilities and designs are so vastly different. Differentiation is key in this market, just open the platforms with nice and open development technologies, it is enough!

One big trend we can foresee also is that the platform vendors have no more software complex, and when you look at the list above, nearly all the initiative are coming from OEM, and not really from pure software companies (notable exceptions: Android and WinMob)….PC based paradigm seems soooo far away!

6 months of 3D in flash in 10 minutes from Galdric on Vimeo

Found here: http://www.hebiflux.com/blog/2008/06/12/video-6-mois-de-3d-en-flash/

Seen from the great usability/UE/User experience maintained by TAT:TAT » The Astonishing UI-blog

A Tactile feedback … made of fresh air! (Ok, ultrasound)

Stunning….