Archive Index (1999-2012) | 2013-current at | About  

Article: Executive Interview — Bill Weinberg of the Mobile Linux Initiative

Dec 20, 2005 — by LinuxDevices Staff — from the LinuxDevices Archive

ForewordLinuxDevices recently caught up with OSDL spokesperson Bill Weinberg, to discuss the organization's new Mobile Linux Initiative (MLI). Weinberg, who serves as the OSDL's representative to the MLI, filled us in on the initiative's early progress, strategy, and… tactics for the year ahead. Enjoy . . . !

Q1 — Can you bring us up to speed on what the MLI (Mobile Linux Initiative) has been up to recently?

A1 — Sure. For starters, we've recently gained a couple of new members, who joined the OSDL with the particular focus of being involved in the Mobile Linux Initiative. These initially included Motorola and PalmSource, and we've since added Mizi. Also, Siemens just joined, explicitly in order to join the Carrier Grade Linux and MLI working groups.

We just had a meeting in Tokyo — we had decided as a body to meet in Tokyo after our launch meeting in Beijing, for a couple of reasons. We wanted to engage with new members and non-members, to gather requirements, and do some additional recruiting, with special emphasis on the ecosystem around DoCoMo.

So, in addition to the membership attending, we had observer members from two categories. We had observers who were members of OSDL, but not participants yet in MLI, and then we had some outside participants, who were there as observers. We structured the meeting in such a way that we had a day and a half that was open to the outside, and then a day of exclusively membership meeting.

We had I believe 25 attendees from a variety of companies, and within the membership of the OSDL, there were people there from NEC and NTT as observers, as well as people like MontaVista, PalmSource, and Intel. That proved to be very fertile, very good discussion. We accomplished a couple of different things.

At the launch meeting, we had set up our initial working groups. We appointed in our chairpersons, set some milestones — such as when we would put our governance in place — and began talking about our technical agenda.

In Tokyo, as a followup, we brought some new people in — we had three new members — and we really layed out our agendas for 2006. I can give you some visibility into that.

We set about doing things like gap analysis, so we had presentations on competing mobile operating systems, both current and legacy. And we agreed on the following mission. The highlevel mission we decided upon is that we're stating that we're accelerating Linux adoption in the mobile space, through a number of strategies.

For example, we're identifying and addressing both technical and non-technical requirements. We feel the non-technical ones are often as important as the technical ones, especially in terms of acceptance, how carriers and operators view Linux in the handset, and how Linux interacts with regulatory statutes in place, such as FCC requirements.

The second strategy we have, is that we're very much aware that open source speaks with code, and the members we've recruited are very much interested in joining that dialog. So, our second agenda is to create and foster implementations, in open source. We're not specifying the scope of those implementations, but rather that we are committed to working and creating technology.

On the non-technical side, another strategy is to advocate and explain industry needs back to the open source community in general, and to the kernel developer community in particular. We think we need to be a bridge in that area, because there is something of a disconnect there. The kernel community has its own set of interests, that are not aligned — and may never be entirely aligned. But, we would like them to be complementary to what we're seeing in the mobile space.

And, we have some marketing goals, like promoting mobile Linux, educating carriers and others about open source benefits, and clarifying legal and regulatory issues for the industry as a whole around mobile phones, with a particular emphasis on Linux and open source.

Linux's share of the smartphone market expanded from 13.7 to 25 percent between Q1 and Q2 of 2005, according to Gartner data

I don't know if you're familiar with the kinds of things that are issues. Certainly there are requirements from the FCC that radio devices not be able to be hacked — for example to change the range characteristics — after they've been certified. The main concern about hacking, really, is changing [a Linux phone's] power. Some of it is safety, some of it is how the FCC likes to see the radio spectrum used.

There are certainly other areas. There needs to be some alignment and standard methods for harmonizing the desire on one hand of having a completely open phone, and on the other hand of being able to build devices that conform to regulatory regimes.

The final strategy goal under this top level is to enable and foster what we call a “pre-platform developer ecosystem.” We settled on some terms that we'll be using a lot.

“Pre-platform” is really working with handset manufacturers to define standardized handsets as platforms both for cultivating a developer community, and for deployment to carriers and operators.

“Post-platform” is the development of applications for fully developed platforms/devices. [It is] enabling aftermarket development and helping ISVs. Part of our agenda will have pre-platform and post-platform flavors. However, we decided to defer post-platform or after-market activities probably for a year, and focus on pre-platform in this initial stage.

We think in the short term, we can help most by helping to define a stable platform, [solving] pre-platform problems. [For example, defining], well-understood, standard methods of doing power management, profiling and scalability, and other low-level things. In the pre-platform area, we realize there's work to be done in the area of storage, in particular optimizing filesystems and flash filesystems. Helping to standardize on a multimedia framework. At a higher level, how codecs are built and deployed, how you deal with IP (intellectual property) issues. Addressing sort of the last asymptotic millimeters in some cases of performance issues. Certainly looking again at real-time, boot time, and, we put footprint under performance. We're making a study of development tools, and where the gaps are in development tools for the mobile space. We have a requirement for helping to make bluetooth and WiFi and other networking stacks more readily available, to lower the barriers of entry — and we think that will extend to telephony stacks as well. Certainly optimizing memory management for challenges like heterogeneous and non-contiguous memory, which while doable, is not a standard, supported item in most Linux implementations. And also, dealing with some of the vaguaries of out-of-memory (OOM) handling, and how the reapers work.

The last technical area we're going to address is security, but we postponed the parameters of that discussion until our next meetings.

Q2 — Did you say “reapers”?

A2 — A reaper is a daemon or a kernel functionality that reaches out and terminates processes, according to some criteria, to free up memory. I first encountered some of the side-effects of the reaper at MontaVista, when we were running some tests on the 2.4 kernel. Tests would spontaneously die overnight on small memory systems, and it turned out that they were running out of memory, and the reaper was being invoked without any particular criteria to what would get terminated. Learning to configure the reaper on Linux systems — most developers don't even know it's there. So that was a fertile discussion, all by itself.

So, those are the technical areas that we think we're going to be looking at. Right now, we're in gap bridging mode. We're trying to be very conservative with regard to defining aggressively a “big tent” platform. We're very wary of overly ambitious efforts, and we're much more interested in making the phone space — for Linux — more viable.

Q3 — Can you explain what you mean by “big tent”?

A3 — I think there've been a lot of ambitious efforts at standardizing portions of Linux for embedded applications that have been less successful than they might be because they bit off more than they were able to chew at the time. And actually doing work in incremental quantities is much more consistent with open source philosophy and practices.

On the non-technical front, we made some decisions to focus in four areas. One is legal. One is working with silicon providers to get alignment on how they deploy Linux on their evaluation platforms. Certainly, addressing fragmentation among versions of Linux for the handset space is something of interest to us. We think there's a lot of work going on below the value line, and we'd like to establish some norms for where people should cooperate entirely in the open source fashion, and where we think they should be innovating and differentiating.

Right now, Linux platforms are probably differentiated too far down the stack. The outcome of that is that when you look at the two dozen or so Linux phones on the market (story), they're relatively un-related, in that they run Linux, but they have little else in common. And that does no favors to ISVs or other developers looking at deploying new applications or services on them.

And, I think the greatest service we'll do for the community is work with — work with the mainstream community — to help bring mobile-enabled technologies — many of which we've just mentioned — into the mainstream kernel, as opposed to having them live their lives out as patches or architecture-specific items. A great example of where that fragmentation is really easy to see is in power management. The mainstream Linux community is focused, really, on notebook power management issues, and we think we'd all benefit by combining our efforts a little more, and moving to a more generalized paradigm.

Q4 — In that regard, is CELF (Consumer Electronics Linux Forum) still really active? Or is it kind of dormant? Is MLI taking over that initiative?

A4 — I'd rather you address that question to Tim Bird or Scott Smyers (of CELF). We're having very good conversations with Sony and others inside CELF, and the OSDL and CELF have 17 members in common. So we have a very strong dialog. I think we're taking a different approach than MPP is, and we'll have different results. You might want to ask Tim or Scott about some upcoming publications or announcements they're expecting to make.

Q5 — What is the MPP?

A5 — Mobile Phone Profile. It's the output of a working subgroup inside CELF.

Q6 — What about the LiPS (Linux Phone Standards) Forum? Have you had the kiss of LiPS?

A6 — [Laughs] We actually have a large number of members in common. We recently had a very fruitful meeting with some of the representatives from LiPS, like Jaluna. We have membership in common, with PalmSource and people like that. And my impression is that LiPS is really working on a higher level up the stack — with regard to service delivery — and they're defining APIs at that level. They've been querying us as to what are we are defining at a lower level, and asking if they can implicate whatever we do in their specification. And our answer has been to date that it's just a bit too early to say, but that we're eager to cooperate with them. And we are.

Q7 — It seems like it's certainly to the advantage of Linux on mobile devices that all three of these efforts — which are the only real ones going on, other than efforts by individual vendors that are all members of these groups anyway — need to get into sync, and avoid redundancy…

A7 — We're painfully aware of that. We're trying to manage our approach to be the most pragmatic possible, which is why I've said we're not in a rush to define a platform. We're more eager to launch resources and fill up gaps. We have the resources, and the members of MLI joined with the specific agenda of working to advance technology.

I think LiPS has something that it has to resolve, in that it is at such a high level, that it's almost coincidentally Linux-based. It could just as easily be iPS, or WiPS, or SiPS. The companies in it have a distinct interest in Linux, and that's why it's LiPS.

Q8 — Is MLI going to have any events out here in the US?

A8 — We're going to have a face-to-face meeting in the Silicon Valley in Q1, and they'll probably be some portion of that that'll be open to outsiders. We're in the process of deciding the exact date and venue there.

And, we are participating, presenting, and otherwise supporting LinuxWorld in Boston. We're likely to have some involvement at 3GSM. Just this last week, we've been in very deep discussion on those two events in particular. I'm presenting at Boston on Mobile Linux, and I'm also presenting at the O'Reilly Emerging Telephony Conference, on open phones, at the end of January.

And there's other folks; Ibrahim (Haddar, MLI manager) is doing some presentation work, and other people. For example, I just had a meeting with PalmSource yesterday. We talked about their media calendars, and there's a lot of things they're going to be doing. So, you're going to hear a lot more about MLI in the first and second quarter. We don't have plans right now to stage our own conference, or anything. But there will be lots of activity.

The OSDL has two multi-initiative face-to-face meetings each year, like the one we just had in Beijing. And then each initiative can make it's own face-to-face meetings, in addition to its phone meetings. As is normal with a new initiative, we're having much more frequent face-to-faces than the more mature initiatives, to get to know one another, and to integrate new members. In the first quarter, we'll be having our elections for governance, at the end of January.

Q9 — Do you personally have an official affiliation with the MLI, or are you working today as a spokesperson for the OSDL?

A9 — I'm the spokesperson for the OSDL in general, but my particular affiliation with the MLI is that I am the OSDL representative to MLI as a body. [The OSDL always has] one rep and one alternate rep to any initiative, where we participate as though we are a member, except that we don't vote, except to break ties.

In addition, MLI is staffed by Ibrahim Haddad, who is the initiative manager. He handles the logistics and the day-to-day workings of the initiative.

I've been assigned as part of my duties to be the OSDL representative to this initiative, and to help drive its inception, because of my background and domain knowledge.

Q10 — The name “Mobile Linux Initiative” is very similar to “Mobilinux,” MontaVista's Linux phone ecosystem. People are sometimes wary of organizations that appear to be too closely affiliated with a single vendor, as we saw with Eclipse and IBM. Has there been any pushback on the name?

A10 — MontaVista is a rank-and-file member. The way we've discussed that very issue within MLI is that the M ought not to be Motorola or MontaVista, and the I ought not to be Intel. And, we're certainly recruiting to make that the case.

MontaVista's Mobilinux is spelled differently, and is a proprietary third-party program initiative. I'm very familiar with their recent announcements. Did you see they just put up a whitepaper? The whitepaper is interesting, because it talks about the integration of real third-party software, and how it works on a phone. It was interesting to me, because I'm talking to the same people that are in the whitepaper about possible involvement [with the MLI].

So, I don't think the coincidence in names is a big issue. And in fact, we had all sorts of debates about what the initiative should be called, but we eventually decided not to worry about it, because we had our launch to get on with. MontaVista's a member, and we're glad to have them. But in the same kind of role, we also have Wind River, and then if you add middleware and applications, we have Mizi and Trolltech and PalmSource.

Q11 — Do you feel you have succeeded in recruiting most of the key members who are active in this space?

A11 — There are two areas where we're emphasizing our recruiting short-term. One is getting the other key silicon players to join, and the other is recruiting more handset manufacturers. We're well into dialogs with the appropriate suspects.

Q12 — How many members are there currently?

A12 — Nine, including (in alphabetical order) BT, Intel, Mizi, MontaVista, Motorola, PalmSource, Siemens, Trolltech, and Wind River.

Organizationally, we've moved pretty quickly. We did our gap analysis, we looked at our deliverables and planning for next year, we've actually done a lot of outbound marketing already — you guys were very generous to us in talking about MLI. We have something like 150 press mentions, and we're in the process of placing articles right now. I'm writing a couple, and others are writing a couple, that you'll see coming out soon.

Q13 — When you mentioned security, you didn't mention the letters DRM (digital rights management) — is that implicit within the definition of security?

A13 — We left the exact definition of security open for basic discussion, for two reasons. The first is, we've identified something like five areas of security that relate to portable devices. There's physical security, access, intrusion detection, content protection, and (data) stream security. Content protection is where DRM would fall. I once wrote an article about this. We're trying to make some decisions based on what the membership wants to focus on. Before the next major meeting, we've decided to liaise with the OSDL's special interest group focused on security that cuts across initiatives.

Q14 — Getting back to the notion that Linux phone users might illegally modify their devices — is that an actual possibility?

A14 — Today, it's not. You guys reported on Harold Welte's OpenEZX site (story). He sort of despairs that he can't penetrate beyond a certain point, and of course you would be voiding your warranty if you began to do that.

What I see in the foreseeable future on phones is that there are open platforms for user applications, and some addition of system code, but not all the way down the stack to the driver and kernel. And that's certainly going to be the desire of the handset manufacturers themselves, who want to be able to define a well-delineated device, and the requirements — some of which are stipulated by regulatory bodies — that come to them directly from their customers, the carriers and operators.

Q15 — Does maintaining a separation of openness for users and security for operators represent a large technical burden for Linux phone vendors?

A15 — Well, Linux is particularly good at offering layered access, so enabling the development of userspace-only applications of some sort is well within the technology's capabilities. Today, what I like to call the “can opener” for handheld devices has really only been Java, and then certain sandboxed APIs like BREW, all of which are relatively constraining, and have had issues with performance.

One of the things we're trying to do is help ISVs (independent software vendors). I was in an interview with a game manufacturer that has to do as many as 140 different spins of a single game for all the different phone permutations. Our goal is not to create 141.

We need to find some unifying principles and make it easier for ISVs. But I think the more immediate beneficiaries in this space will be distributed enterprise application developers, who'll want want to move enterprise application client components onto these intelligent phones.

Q16 — And not be forced to use Java or another managed environment?

A16 — Right. And they'll be able to make relatively arbitrary, scalable decisions about where the functionality lives in the network — on the handset, points of presence out to other servers, and back. The kind of work we're doing, and some of the other organizations are doing, will benefit all kinds of ISVs.

Q17 — What about on the UI level? Certainly one problem is that the UI has to be really simple and intuitive. Will there be any kind of unification of the UI for the Linux phone space, at least as an option?

A17 — The current state of affairs is somewhat similar to the desktop, in that there's a couple of dominant [UI stacks]. And of course if you look out at the phone world, there's a handful of dominant interface paradigms. Multiply one against the other, and you end up with a lot of fragmentation.

We haven't broached that topic yet with our members who are in the UI business. Certainly there is a different approach taken by PalmSource versus Mizi versus Trolltech. We have to leave it to the membership to decide what is above and what is below the value line. So that's a topic of discussion.

Q18 — I guess you'd have to add Motorola into the mix…

A18 — If you look at Motorola, they've got Windows phones, Linux phones, and Symbian phones. But the Linux phones, to the best of my knowledge, use Trolltech.

Q19 — Yes, but they don't use Qtopia. They use the Qt/embedded framework, but they have their own UI on top of that.

A19 — Right. They have their own UI traditions.

I guess [the UI] probably persists above the value line, and more of them may become more popular.

One of the areas of customization may be that someone in the value chain would be able to easily customize or even swap out pieces of the UI. Given that that's part of most vendors' branding and differentiation, I doubt that will be the first thing to be opened. What's more likely to be opened is new ways to deploy applications.

Q20 — In terms of competitiveness, that all sounds good, and I imagine some operators and device makers will be attracted to that. It'll be interesting to see how it plays out in the purchasing community — enterprise, or even consumer — in terms of competing with something like Windows Mobile, where every single Windows Mobile phone is the same in terms of how it works.

A20 — That's a consideration of ours. Linux has been growing in marketshare on smartphones, and is well ahead of Windows. It's just that we want it to stay that way, and as application delivery becomes more important, that's an area where Windows has a lot of potential to gain back marketshare. Probably at the expense of Symbian, but certainly, it won't help Linux build volume, either.

Q21 — Obviously there are advantages to Mobile Linux, and obviously there are perceived advantages — and they're significant ones — to Windows Mobile…

A21 — Yes, the fact that Windows Mobile is a ubiquitous, well-known, out-of-the-box solution, that has probably more features in the box, and can accelerate getting to prototype faster.

Q22 — But also, if you separate what Windows Mobile has to offer — and I've looked at it closely enough to know there are a lot of aspects that are advantageous — one important one is the ability it has to connect with enterprise applications. And also, it's beginning to plug into desktop Windows applications and data and functionality. Those seem to be areas that, in a sense, could be considered to be below the value line, because they're fundamental capabilities. Such as email push — an interesting one — or other kinds of synchronization of data from phones and PCs. It seems like that kind of functionality needs to be available for any form of mobile Linux.

A22 — The synchronization issues are becoming increasingly standardized. Certainly, the ability to support what's coming out of OMA (Open Mobile Alliance) is very important. Motorola phones do that, and I've talked to, for example, the folks at Funambol about their client-side approach to Linux, and they're very interested in MLI. They may not be ideal members in the short-term, but we've had a very fruitful dialog with them.

I think that it'll come down to the quality of OMA clients, in terms of their competitiveness. And I think there, Linux will have an advantage, because the trend is always to do things using the most standards-based APIs, standards-based technologies. Whereas Microsoft has a tendency to go off and do things their own way. Even if they're very attractive, it creates barriers to entry, and it does, let's say, distort the value line.

Q23 — Thanks very much for your time, and please help us keep our readers informed about goings-on at the MLI

A23 — My pleasure, and we will.

About the Interviewee — Bill Weinberg brings over 18 years of open systems, embedded, and other IT experience to his role as Open Source Architecture Specialist and Linux Evangelist at the Open Source Development Labs, where in addition to the Mobile Linux Initiative, he participates in OSDL initiatives for Carrier-Grade, Data Center and Desktop Linux. Prior to the OSDL, Bill was a founding team-member at MontaVista Software, and helped establish Linux as a favored platform for next-generation intelligent device development. In his extensive and varied career, Bill also worked at Lynx Real-Time Systems, Acer Computer, and Microtec Research. Today Bill is known for his writing and speaking on topics that include Open Source licensing, international adoption of Linux, embedded/real-time computing, application porting/migration, and Linux-based consumer electronics and handheld applications.

This article was originally published on and has been donated to the open source community by QuinStreet Inc. Please visit for up-to-date news and articles about Linux and open source.

Comments are closed.