westmeadboy
榜眼
Mike - you might be interested in the latest update to the Android NDK: http://developer.android.com/sdk/ndk/index.html
Google Android
- Thread starter westmeadboy
- Start date
Though it's a shame that they didn't add these Native Activities last summer - it would have taken a good bit of engineering work to get a native-code UI framework like Qt up and running, but it would have probably been less than the amount of work we've invested in making our code base accessible from Java, so had they been available sooner we would have probably tried to do the whole thing with the NDK.
Interview with Andy Rubin with Walt Mossberg and Kira Swisher at the D: Dive Into Mobile conference.
http://mobilized.allthingsd.com/2010120 ... o-android/
Google’s Andy Rubin Gives a Flash of Tablet Future
by Ina Fried
Posted on December 6, 2010 at 6:33 PM PT
http://mobilized.allthingsd.com/2010120 ... o-android/
Google’s Andy Rubin Gives a Flash of Tablet Future
by Ina Fried
Posted on December 6, 2010 at 6:33 PM PT
Thanks - the comment about "adding multiple APIs" is a good sign - suggests that they're not totally redesigning the Android UI frameworks for tablets but simply expanding them, which Apple did but HP/Palm didn't do (and if you consider the N8xx series to be "tablets" then neither did Nokia).
Sarevok said:
Odds are good, but some recent experiences have me even more worried about phone-specific bugs / problems, so I really wouldn't buy any particular Android phone expecting to run Pleco on it yet; the only phone we can guarantee will run Pleco correctly right now is an iPhone.
megaheld said:
We'll announce it, yes. End of year early test version didn't happen for basically two reasons: 1) Android 2.3, which took forever but has enough exciting stuff in it to be worth waiting for / not rushing something out for 2.2, and 2) OCR, which was far more successful than we anticipated but needs a lot of work to make it less jittery / more still-image-friendly - i.e., we had a reason to push it back and something else to do with the time if we did.
Just had our first successful test run with live OCR on Android (after taking delivery of the code library only yesterday). This was on an HTC Desire, essentially identical to a Nexus One.
The system seems to be functional - we can pull data out of the video frame, feed it to OCR, and get the correctly-recognized characters - but the Android camera API is horrible; in theory it ought to do almost everything we need, but in practice it does hardly anything for us and is riddled with manufacturer-specific bugs / quirks (confirmed in several developer forums). We can't even get it to reliably draw a preview of the camera output, for example, so we're basically going to have to pull raw video data off of the camera and do the rendering ourselves - not that that's all that difficult, but it really is the sort of thing that the OS should do for you.
So live OCR should work on Android, but it's going to be very hardware-dependent: you absolutely won't want to buy any Android phone expecting to run OCR on it until we've actually had a chance to do testing on that exact phone - even if people have had success with other phones from the same manufacturer - and you won't even want to install any significant OS version updates until we've had a chance to test with those new versions. (if you thought our website had too much BOLDFACED WARNING TEXT on it before, just wait...) On the plus side, the fact that people can install Android software freely should let us be pretty efficient about testing / distributing patches for phone-specific problems as they come up.
Performance seems fairly solid, though, a bit slower than an iPhone 4 but the gap should narrow once we get things optimized. The Desire's camera isn't as nice as either compatible iPhone's (doesn't really have a proper macro mode, though its minimum focus distance is fairly decent) but that at least is something that you should be able to test in a store - see how close you can hold a device to a piece of text and have it stay in focus.
The system seems to be functional - we can pull data out of the video frame, feed it to OCR, and get the correctly-recognized characters - but the Android camera API is horrible; in theory it ought to do almost everything we need, but in practice it does hardly anything for us and is riddled with manufacturer-specific bugs / quirks (confirmed in several developer forums). We can't even get it to reliably draw a preview of the camera output, for example, so we're basically going to have to pull raw video data off of the camera and do the rendering ourselves - not that that's all that difficult, but it really is the sort of thing that the OS should do for you.
So live OCR should work on Android, but it's going to be very hardware-dependent: you absolutely won't want to buy any Android phone expecting to run OCR on it until we've actually had a chance to do testing on that exact phone - even if people have had success with other phones from the same manufacturer - and you won't even want to install any significant OS version updates until we've had a chance to test with those new versions. (if you thought our website had too much BOLDFACED WARNING TEXT on it before, just wait...) On the plus side, the fact that people can install Android software freely should let us be pretty efficient about testing / distributing patches for phone-specific problems as they come up.
Performance seems fairly solid, though, a bit slower than an iPhone 4 but the gap should narrow once we get things optimized. The Desire's camera isn't as nice as either compatible iPhone's (doesn't really have a proper macro mode, though its minimum focus distance is fairly decent) but that at least is something that you should be able to test in a store - see how close you can hold a device to a piece of text and have it stay in focus.
numble said:
Haven't actually come up with a good strategy for that yet - right now we're just trying to have one or two from each major manufacturer, that seems to cover most of the severe bugs, but down the line we'll get as specific as we need to (assuming Android sales are good enough to justify buying a given $500 phone for the number of $100 dictionaries we'll be able to sell to its users).
sui.generis
探花
Is there going to be a demo for Android? It seems like that would help most people be sure of whether they can handle it.
There are a few other ways to save on buying up units, especially those that are somewhat similar to units you've personally tested. Either an automated standalone program that runs through a routine to verify Pleco's features will work and uploads a log to you (a la PC sysreqlabs), or maybe offer a reduced fee for users of untested devices you're interested in who install and video themselves running through a checklist of operations (and actually deliver a watchable video to you). Then your cost of verifying certain devices comes down to the cost of refunds. Even if you have multiple testers just to be safe, you could whittle verification down from $500 per device to a fraction of that, and maybe earn some goodwill in the process. Just a thought.
There are a few other ways to save on buying up units, especially those that are somewhat similar to units you've personally tested. Either an automated standalone program that runs through a routine to verify Pleco's features will work and uploads a log to you (a la PC sysreqlabs), or maybe offer a reduced fee for users of untested devices you're interested in who install and video themselves running through a checklist of operations (and actually deliver a watchable video to you). Then your cost of verifying certain devices comes down to the cost of refunds. Even if you have multiple testers just to be safe, you could whittle verification down from $500 per device to a fraction of that, and maybe earn some goodwill in the process. Just a thought.
sui.generis said:
There'll be a demo, certainly, but if someone actually encounters a problem there isn't much we can do without having the device in hand, and given that we might introduce problems in any new update we need to have at least a reasonably good sampling of devices here so that we can verify it on them in advance.
This strategy might make sense for some less common devices, though. The main question is whether it's actually worth even attempting to support them (and being stuck continuing to support them in new updates for people who've bought a previous version to run on them); it may be that the only way we can keep Android development remotely manageable is to have a strictly limited device list - even set a hard limit of how many officially-supported devices we'll have at any given time - and make all other device support unofficial and subject to potentially going away in a new update.
Well, I would be more than happy to test a pre-production version on a Sony-Ericsson Xperia X10! I'd do my best to file detailed reports of any bugs experienced, as well as offer general user-experience feedback.
Is it only the OCR features that will require Android 2.3 or better, or will basic dictionary functions also be unavailable on Android 2.1/2.2?
I just saw your iOS product for the first time today and was completely blown away by the OCR!!! Later, I got my roommate to install it and took a peek at the other functions; Pleco looks fantastic, going way beyond what I have in HanPing. Pleco looks to provide everything that a quality electronic dictionary should, which is more than I can say of any Android app I have seen yet (for Chinese or Korean to English)
I am only one Android user, but I would gladly pay for your dictionary, even if the OCR functions were unavailable until I upgrade to 2.3. I payed $5 for ad-free HanPing, and would not hesitate to pay the same or more for Pleco basic.
I'm not saying you should change your business model, just that there seems to be no competition for what you offer, and that even without the flashier features, you should be able to make a profitable foray into the Android market.
Is it only the OCR features that will require Android 2.3 or better, or will basic dictionary functions also be unavailable on Android 2.1/2.2?
I just saw your iOS product for the first time today and was completely blown away by the OCR!!! Later, I got my roommate to install it and took a peek at the other functions; Pleco looks fantastic, going way beyond what I have in HanPing. Pleco looks to provide everything that a quality electronic dictionary should, which is more than I can say of any Android app I have seen yet (for Chinese or Korean to English)
I am only one Android user, but I would gladly pay for your dictionary, even if the OCR functions were unavailable until I upgrade to 2.3. I payed $5 for ad-free HanPing, and would not hesitate to pay the same or more for Pleco basic.
I'm not saying you should change your business model, just that there seems to be no competition for what you offer, and that even without the flashier features, you should be able to make a profitable foray into the Android market.
mandu said:
We're planning a public beta on Android, actually - since we're not constrained by Apple's 100 test device limit there's no reason not to make it available to anyone who wants it.
mandu said:
Not 100% sure yet, but in general I'm inclined to make the whole thing 2.3 only; supporting fewer OS versions makes compatibility testing a lot easier, and there are some basic features in 2.3 (like a systemwide download manager) that it would be very useful for us to build around / assume are always available.
Of course it's still possible that 3.0 might introduce a totally new UI framework that would compel us to make that the minimum rather than testing the whole thing on both - nothing is going to be final in terms of OS requirements until the finished app is released, we could release a dictionary-only beta that works on 2.3 but limit the finished app to 3.0 or 3.1.
We've wasted way too much time in the past on ensuring backwards-compatibility on other OSes, so we want to start our Android version out with the absolute minimum number of official devices / OSes supported so that we don't have to spend a lot of time supporting devices / OSes that are seeing their installed base dwindle. Our high prices are actually a big part of the problem on this - people can live with the occasional bug or glitch in a $1 or $2 or $5 app, but if they're shelling out $100 they have every right to expect it to run perfectly on their device, and we can only do that level of compatibility testing / optimization on a small number of devices and OSes.
So my guess is that we'll start out with a list of something like 5 or 6 officially supported devices, all relatively new ones running whatever the latest version of Android is (hopefully including at least one that's widely available in China), then gradually expand that to a list of 12 or so and maintain the list at roughly that size from that point forward, retiring old devices and introducing new ones as appropriate. Other devices might work too, but there'd be no guarantee that we'd fix things for you if they didn't - if you wanted to buy an Android device to run Pleco, we could only promise that that would work if it was on our list. Of course if the Android version is a smash hit we might be able to do more testing and support more devices, but 6-12 would probably be our baseline.
mikelove said:
I don't know what kind of time-line you expect for an Android version, but of the large user-base as of Dec 2010, a majority should be able to upgrade to 2.3 in time, barring difficult manufacturers (shakes fist at Sony). I have gotten the impression that 3.0 will be such a major upgrade that current devices will largely be shut-out. Rumours have swirled of 512MB RAM minimum, possibly >800x480 minimum resolution, etc.
If you can target the "high-end" Android phone of today (1Ghz, 256MB, 800x480/854x480, and 2.3), you should have a good base of potential customers. If I were able to get started with Pleco on my X10 (assuming 2.3 arrives for me), and transfer the licence along with my google account when I eventually upgrade to a supported 3.0 device, you would have my money.
I would expect better penetration in China for current "high-end" Android devices than for dual-core,512+MB, 1280x720 monsters that will cost a fortune.
Unless google intends to release a 3.0-lite for everybody who got on board in 2010, I would try to target Gingerbread to maximize your potential market.
sui.generis
探花
mandu said:
I bet xda puts it on a G1 within a month of release. I think specs will be less of a barrier to most upgrades than manufacturer/carrier choice, and there were those same rumors about gingerbread. The really popular android phones have been the fairly high end versions anyway, with cortexes and snapdragons and hummingbirds (oh my!) and WVGA. Besides, I don't think Google's culture is compatible with going Vista on us. I'm not even convinced by the rumor that 3rd party UIs are really out, though I'd be thrilled if baking them into the OS was out and they had to be added on from the marketplace or some such.
That said, I tend to agree with your conclusion about targeting gingerbread initially (not so much by including ginger devices on their updated device list, but by having the initial stable pleco functional without 3.0 hooks). Just because (if I'm right) most current models can handle 3.0 doesn't mean they'll get the chance officially. Most users don't download questionable releases from XDA, they wait on their carrier. Even if 3.0 devices sell like hotcakes, there are millions of droids, galaxies and various HTCs that probably will stay afloat on the great material river for a few years more and there's an incentive for carriers/manufacturers to leverage honeycomb to encourage upgrades.
I doubt it will affect me much, personally. I'm about to start dual-booting android on my TP2, so I can continue to use the winmo version until HTC wises up and makes a TP3 (Pro 7) android-flavored. And by then, I'm sure 3.x will be standard. Just a recommendation because I want Pleco to succeed and prosper.
mikelove said:
That sounds fair to me.
sui.generis said:
1000% agreement!!!
I hadn't heard about limitations on customization, but that could only help Android at this point. Any moves google makes in that direction will help projects like Pleco that rely on accessing the HW in a consistent fashion. Google should prevent them from routing the camera through custom software (Sony) or otherwise interfering with Android.
In Canada, it's nearly impossible to buy a pure Android phone.
Some details just came out about Amazon's app store (perhaps soon to be powering a future Android-based Kindle):
http://techcrunch.com/2011/01/05/amazon-android-app-store-2/
Unfortunately the T&C's make it completely impossible to release Pleco through their store - even if we didn't mind the vetting / approval process, not being able to control prices means we can't guarantee our minimum pricing commitments with some dictionaries or guarantee that we'll be making enough money to cover the minimum royalty for others; ironically (or perhaps deliberately) for a company that's known for e-books, Amazon's store makes it almost impossible to sell apps containing licensed content.
Fortunately, since we're only planning to support a limited subset of the available Android devices anyway, if some devices (say that future Kindle) come with Amazon's store built in and no way to run other apps we can simply support competing devices instead - the people who are most interested in running Pleco on Android are unlikely to be satisfied with a locked-down device anyway.
http://techcrunch.com/2011/01/05/amazon-android-app-store-2/
Unfortunately the T&C's make it completely impossible to release Pleco through their store - even if we didn't mind the vetting / approval process, not being able to control prices means we can't guarantee our minimum pricing commitments with some dictionaries or guarantee that we'll be making enough money to cover the minimum royalty for others; ironically (or perhaps deliberately) for a company that's known for e-books, Amazon's store makes it almost impossible to sell apps containing licensed content.
Fortunately, since we're only planning to support a limited subset of the available Android devices anyway, if some devices (say that future Kindle) come with Amazon's store built in and no way to run other apps we can simply support competing devices instead - the people who are most interested in running Pleco on Android are unlikely to be satisfied with a locked-down device anyway.