the prognosis for wm 7

[mikelove]

It doesn't, the running-in-the-background thing is about the only type of multitasking we'd ever need unless we developed an IME (and that requires OS-level support in other ways anyway). And it's the only kind we'd get on iPhone too unless we start supporting background audio playback (though that's certainly a possibility in a future version if we add some new specific audio learning features).

That being said, with HP buying Palm I think the prospects for Microsoft in the mobile space have just gotten a lot bleaker - webOS does basically the same thing as Windows Phone 7, offers an web-code-based API for regular apps and a native-code one for games and processor-intensive apps, but it does it using nice open standards, is if anything even more accommodating of unsigned / un-approved apps than Android (while WP7 seems unlikely to run them at all without hacked firmware), will have a year and a half of extra maturity / mindshare / feature additions by the time WP7 hits, and will have a hardware manufacturer putting it front and center - most of Microsoft's other launch partners are very much in the Android camp and aren't really committed to WP7 anyway. If anybody other than Apple, Google, or hanging-on-in-spite-of-their-continued-mediocrity RIM is going to carve out a corner of the smartphone OS market, it's probably HP / Palm. The only thing Microsoft really has going for them in this battle is Office, but I have a hard time thinking DataViz or QuickOffice or one of those guys won't have a very solid webOS version out by the time WP7 hits.

 

[character]

If anybody other than Apple, Google, or hanging-on-in-spite-of-their-continued-mediocrity RIM is going to carve out a corner of the smartphone OS market, it's probably HP / Palm.

I'd put my money on Nokia first. What is the only HP OS to grab mindshare? Their calculator OS. I'm not sure it's in their DNA to attract developers. OTOH, HP is well positioned to capture customers as their WM6 phones wear out. Personally I think Microsoft will realize they are losing business customers by not having a real MS smartphone OS and come out with something, maybe a stripped down Windows 7 on a 1GHz+ processor (intel has been trying to get Atom into smartphones).

The only thing Microsoft really has going for them in this battle is Office, but I have a hard time thinking DataViz or QuickOffice or one of those guys won't have a very solid webOS version out by the time WP7 hits.

Also worth noting Office formats are now standards, so MS can't easily prevent others from reading/writing Office files.

 

[mikelove]

The difference is that Nokia has a very long track record of failing miserably at smartphone OSes - certainly the Qt stuff they're working on for Symbian^something-or-other looks promising, but so have a lot of the other things they've tried unsuccessfully over the years. Palm on the other hand has been badly mismanaged but has a proven track record for delivering really nice software; given their dire financial situation over the last year, the fact that webOS has attracted even as much attention as it has is a testament to their continued software development skills.

Nokia has also been an abject failure at selling phones in the US, which is where an awful lot of smartphone software developers live and make much of their money (heck, on Android it's almost the only game in town), while HP is extremely talented at pushing product here - walk into the PC section of any Best Buy and you'll see a ton of HP logos amidst the rows of commodity laptops / netbooks / etc. When HP launches their first post-merger webOS phone, you're not going to be able to go into an electronics or cell phone store in the US without walking past a big glowing cheerfully-staffed display trying to persuade you to buy one.

Also worth noting Office formats are now standards, so MS can't easily prevent others from reading/writing Office files.

Wish that were true - the OOXML spec (rammed down the throats of international standards bodies amidst very considerable opposition) is a buggy, thousands-of-pages-long monster that even Microsoft's own products haven't fully implemented yet. In fact even the upcoming Office 2010 isn't entirely compatible with OOXML.

 

[sui.generis]

What is the only HP OS to grab mindshare? Their calculator OS. I'm not sure it's in their DNA to attract developers.

HP just got a DNA transplant. A very impressive one.

 

[gato]

HP-UX was popular, as popular as workstations can be, in the 1990s.

 

[character]

HP-UX was popular, as popular as workstations can be, in the 1990s.

Really? I thought Solaris was the clear leader, but perhaps it depends on where you were/what you were doing.

The difference is that Nokia has a very long track record of failing miserably at smartphone OSes - certainly the Qt stuff they're working on for Symbian^something-or-other looks promising, but so have a lot of the other things they've tried unsuccessfully over the years.

I guess we'll see. Maemo looks good, it just needs a better phone and more polish.

WRT the US market, it's probably going to be less than half the market for apps fairly soon. Add in that only men seem to install a lot of apps, and I'm not sure a big body of developers will be that key going forward.

 

[gato]

I thought Solaris was the clear leader, but perhaps it depends on where you were/what you were doing.

Maybe. I was an intern at HP Labs.... Heheh.

 

[mikelove]

Two interesting news stories on this front today, actually. Intel Atom chips are getting lower power requirements, increasing the odds of a desktop-Windows-based smartphone (as I said in another thread, I don't see a lot of Android manufacturers moving to Atom given that ARM is still a generation or two ahead of them power / speed-wise), and Microsoft is apparently working towards that very thing with their "Menlo" project.

This might actually go a long way to explain why WP7 only runs managed-code Silverlight / XNA apps; they're breaking compatibility now so that when their real mobile platform move comes along, the long-awaited desktop-mobile convergence device (an Atom-powered smartphone that runs desktop Windows 7 when you dock it / plug it into a monitor but instantly switches into mobile mode when you take it out), apps developed for WP7 will work on it seamlessly.

 

[gato]

Intel's probably never going to catsch up in this market. They still two years behind. They must be regretting their sale fo the XScale business, which is now Marvell. It could have been them making the announcement below. Intel has made some wrong turns over the years, just like Microsoft.

http://mobile.pcmag.com/device2/article ... 092,00.asp
Marvell Launches High-End Smartphone Processor
By: Sascha Segan
02.11.2010

In advance of next week's Mobile World Congress trade show, Marvell on Thursday launched another mobile processor that directly competes with Qualcomm's Snapdragon chipset for smartphones and smartbooks.

 

[mikelove]

Oh I don't expect Intel to catch up - heck, the X86 instruction set alone will hold them back a bit (they worked around that on desktops by dynamically recompiling programs into a proprietary RISC instruction set on-chip, but I don't believe they can do that yet on Atom) - but particularly as processors continue to constitute a smaller and smaller share of mobile device power consumption (since it's easier to cut power consumption from them than from, say, a screen backlight), they may come out with an Atom chip that's "good enough" to put in a smartphone. The Palm Pre / Pixi get absolutely terrible battery life relative to their peers but people still buy them for the superior OS - give me a smartphone that I can plug into a monitor / keyboard and use as a fully functional Windows PC and I'll gladly put up with having to remember to recharge it every night.

 

[RobRedbeard]

This might actually go a long way to explain why WP7 only runs managed-code Silverlight / XNA apps; they're breaking compatibility now so that when their real mobile platform move comes along, the long-awaited desktop-mobile convergence device (an Atom-powered smartphone that runs desktop Windows 7 when you dock it / plug it into a monitor but instantly switches into mobile mode when you take it out), apps developed for WP7 will work on it seamlessly.

That is a big part of what is going on right now. Hardware independence has become very important in a market that can make a 90 degree turn with the release of one product. It's going to be interesting to see how Apple deals with this as their "devices" and "computers" continue to moonlight in each others' markets.

Back to managed code, the move also has a lot to do with security and software homogeneity. Windows Mobile was a platform marked by the ability for OEM and development houses to more or less use it as a "clay" and mold it into whatever they wanted. With Windows Phone, that has changed completely. It sucks for tweakers and those of us who like to code close to metal, however, in the long run centralized resource management makes sense for software security. It also means the ability to more reliably implement high level functionality across applications.

Funny thing is Core processors still beat the Atom on power / performance on the desktop, which only leaves two screen. If mobile isn't one of them, and HTPC is still a niche market, then what exactly is Intel's goal with Atom (source: http://www.tomshardware.com/reviews/int ... ,2069.html).

Rob

 

[mikelove]

Apple I think has decided that they can avoid managed code because they control all of their own hardware and hence can prepare / plan carefully for architecture changes. Plus, thanks to the X86-based iPhone simulator's prominent use in in iPhone software development, they already know that basically any iPhone app will run quite happily on an X86-based iPhone OS platform with little more than a recompile if they choose to go that route in the future. I'd say within another year or two there'll probably be a capacitive-touchscreen mini MacBook that runs OS X but also plays nicely with recompiled iPad apps, and a bit further down the line we may see a converged, dockable, OSX-running iPhone.

Good thing for us, anyway - if Apple were as unfriendly to native code as Google and Microsoft, we might very well have ended up doing desktop and web-based apps and nothing else post-WM.

 

[RobRedbeard]

Good thing for us, anyway - if Apple were as unfriendly to native code as Google and Microsoft, we might very well have ended up doing desktop and web-based apps and nothing else post-WM.

Well put, although I must say I draw into question how necessary native code is for Pleco given the power of today's mobile devices. I realize that you must be doing a tremendous amount of your own memory management given you're using your own engine. Have you created a new model for your engine yet in a managed code system? While performance would be reduced, it should take far less time to code. I get the feeling from your other posts that you'd probably move to an installed sql package. I'm just trying to contemplate how this would be done directly in managed code without doing that.

Rob

 

[mikelove]

Oh, the performance for database searches at least doesn't worry me - as I've alluded to in the Android thread, the main performance bottleneck for our database engine is I/O. Even before we ARM-optimized it, our current database engine ran just fine on a Palm Tungsten E; i.e., a device with a 144 MHz 2002-vintage ARM processor ran Pleco well in a 68k emulator. It's really just about the code rewriting - our database engine in managed code will probably still be faster than SQLite in native.

But you're wrong about "less time to code" - that would be true if we didn't already have a nice, mature, well-tested base of native code, but since we do, our time to support a new system that allows for native C development is minimal; it literally took about a day to have that engine code from Palm/WM functional on iPhone, the year of development time was basically all spent on iPhone-specific UI and new features. (and even those had a substantial amount of code reuse from native C code on Palm/WM)

 

[RobRedbeard]

But you're wrong about "less time to code" - that would be true if we didn't already have a nice, mature, well-tested base of native code, but since we do, our time to support a new system that allows for native C development is minimal; it literally took about a day to have that engine code from Palm/WM functional on iPhone, the year of development time was basically all spent on iPhone-specific UI and new features. (and even those had a substantial amount of code reuse from native C code on Palm/WM)

I should have been more specific as to what I was comparing. I meant it would take less time to code the engine from scratch in managed code than it took to do so in native code. I didn't mean that it would take less time to code than the port to iPhone, which was essentially a recompile for your engine. Still, the basic interface design from iPhone would be easy to reduplicate on WP7 and or Android if you ever did it. Hell, I'd like to see it on WM6.5. I'm half tempted to just code the iPhone's flash card interface on WM and give it to you.

Rob

 

[mfcb]

I'm half tempted to just code the iPhone's flash card interface on WM and give it to you.

another fan of my API-idea? hehe

 

[mikelove]

The flashcard screen is probably the single easiest screen for us to touch-optimize on WM, actually, at least outside of the free-answer handwriting box (which seems like it would need a total redesign to fit in the available space, or an iPhone-like transparent overlay); rotate the left batch of buttons by 90 degrees and double or triple the size of the three big main ones and we're basically there for self-scored. That and the main dictionary screen both stand a good chance of being redesigned, it's more in the nitty-gittty configuration screens that we're unlikely to go through and painstakingly separate things to be easily tappable without a stylus. Some of the work for this may actually be done anyway for the desktop version - supporting iPhone-like multiline list items, for example.