Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 8 years ago.
Improve this question
I've been intrigued by all the android world since I first learned about it and would like to get my hands dirty developing for it. The question that comes to mind is if it's worth buying the unlocked phones that Android sells directly or not.
Those phones (link) quoting the Android page:
Run and debug your Android™
applications directly on a device.
Modify and rebuild the Android
operating system, and flash it onto a
phone. The Android Dev Phone 1 is
carrier independent, and available for
purchase by any developer registered
with Android Market™.
Please note that this device is
intended for development purposes, and
may not include certain features found
on consumer Android devices.
So will it be worth it to purchase one of those as a tool for app testing as opposed to developing and testing just on Eclipse or one of the other IDEs and emulators.
-Have you tried it, do you own one?
I'm assuming they have the same specs as the HTC Dream and the HTC Magic, since they look exactly the same although they have a 'developers edition' custom black design not that I really care about the design for this particular item.
All comments are welcomed,
Thanks in advance.
Update:
I'll leave it open until tomorrow to see if there are any more answers, then I'll just pick the most voted since it's really a subjective question with no good or bad answer.
It depends what sort of applications you wish to develop. I find that the emulators very accurately reflect how things work on genuine devices; you can seamlessly connect to either an emulator or a dev phone using the command line tools, the Eclipse tools, the debugger etc.
Also, while you can flash your dev phone to a new OS version, HTC often lag behind (e.g. there's still no 2.0 image available) and it's much easier and faster to just use the emulators. The emulators also allow you to create and test with different screen resolutions, whereas the two dev phones available are only "standard" resolution.
I find it's quite rare that I need to use my ADP1 dev phone for development -- my rooted consumer HTC Hero works fine for most of the development I do.. allowing me to pull files from the device etc. Though the only reason I use my Hero rather than an emulator is because I've been working on an app that uses audio recording functionality.
However, where having a physical device would help is where you need to do specific stuff regarding the camera, audio hardware, orientation and compass sensors, GPS, wireless network access and so on. Should you need to connect a debugger to work on hardware-related issues like the above, then you would definitely need a dev phone.
Overall, it's definitely worth buying an Android phone for testing and demonstration purposes, but whether it's a development phone is up to your requirements.
If you plan on developing apps that you intend to put on the Android marketplace, it's absolutely critical to test on real hardware. You can get away with developing on the emulator for quite a while, but at some point, you'll want to use a real device.
That being said, you can use any android phone for development. There are some restrictions on locked devices, but if you're simply developing against the SDK, any phone will work. With android, you can install an apk directly on the phone without special permissions, so the only real advantage to a dev phone is that you can install new roms without having to root the phone.
Personally, I'd hold off on purchasing one of the older dev phones. From what I understand, they only support up to SDK 1.6, whereas the Droid and some of the other new phones are supporting SDK 2.0 ++.
Wait for the release of the Nexus One from G. The latest rumors are that it'll be released on Jan 5th. So it's just a week or so.
I think that you need a real android device whether it's the dev phone or another handset but a real phone is primordial. The emulator is great but you can't get an idea about the execution speed of your app until you use it on a real phone.
As said before there are a lot of rumors about the nexus one so wait and see!
As for which phone to buy (assuming you're going to get one) I think ablerman is right. I'd wait until January to see if there is going to be some new hardware available.
With regards to the more general question of should you buy one, I think it depends on what you're doing. For the most part, the emulators are fine. They can emulate GPS (you can even load KML to simulate a path), SMS, phone calls, etc. They cannot however emulate acclerometer/compass/orientation sensor data and actually will crash (actually I believe it hangs...) if you try to run code that relies on it. Also, it's difficult to actually debug phone-call related functionality without the dev phone.
They're good phones, I've used the Dev phone 1 (the G1/Dream) and it's nice. It also is a bit faster than the emulators and if you're writing something like a game, it would be really good to test it on the actual hardware.
All in all, it just depends on what you're writing. They're definitely fun to play with regardless as you can do pretty much ANYTHING you want on them.
Good luck with the decision!
I've been developing with the emulator since June. I've found it to be a very near substitute for the real deal, and it's easier to switch between handset configurations/versions. However, not knowing how quickly my apps will run is a concern for me.
The reason I've personally held off buying a handset is that 2009 was the wrong year to buy one. I have a feeling 2010 will very much be the long-awaited "Year of the Android".
+1 to Christopher and I will add - the emulators are great but having a physical phone will give you instant access to the Android Market to verify publishing, statistics and user comments. I also believe using your own app on a physical phone will help you to develop a better app. You do not need a development phone - but at least one physical phone - absolutly.
FYI. Belgium is one of the few countries where it is possible to buy any mobile unlocked. Indeed, the Belgian regulators forbid the forced bundling.
One more Pros for buying a real developer phone :
HierarchyViewer does not work on user builds (i.e. with devices
available in stores.) This is for security reasons.
See the original thread
Hierarchy viewer can be very useful if you have problems with layout being slow, although I don't think it would worth buying a real Developer phone only for this.
As some people made workaround for that problem : https://stackoverflow.com/a/7801475/62921.
Related
Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 6 years ago.
Improve this question
I'm in the market for a basic Android device. My main reason for buying one is for learning to develop mobile apps. While the Eclipse IDE offers a virtual Android device for testing apps, it's not the same thing as a real Android device and won't put me on the same wavelength as Android device users. (The most obvious example: Visibility is poorer in a bright outdoor environment than indoors.)
Some questions:
Is a tablet PC the type of device I should get? I get the impression that it's more versatile than other devices. I don't need a smartphone, because my dumbphone works well for me, and I don't even use it that much. And I HATE the idea of being locked into a specific wireless provider. I'm not interested in ebooks, because I still like books on paper better.
Should I get a device with Android 2.3 or 4.0? On the one hand, over 90% of current Android users are using version 2.3.3 or earlier. On the other hand, 4.0 is better and will be gaining market share. (Would buying an Android 2.3 device now and buying an Android 4.0 device later be my best bet?)
In addition to wifi, what else should I look for?
Are there any brands/models I should avoid? I remember hearing that the Packard Bell PCs were the most unreliable. The Yugo was a terrible value. (An old Oldsmobile Cutlass clunker was a better value - at least as reliable but MUCH cheaper to buy.)
Is there any point in buying used? The listings on Ebay and Google Shopping don't show the used tablet PCs to be that much cheaper than new ones, and there are more new tablets than used ones available.
Ebay listings currently (5-23-2012) show over 200 Android 2.3 tablets selling for $55-$90:
http://www.ebay.com/sch/iPads-Tablets-eBook-Readers-/171485/i.html?Operating%2520System=Android%25202%252E3%252Ex%2520Gingerbread&LH_BIN=1&Internet%2520Connectivity=Wi%252DFi%7CWi%252DFi%2520%252B%25203G%7CWi%252DFi%2520%252B%25204G&_nkw=android+2.3&_dmpt=US_Tablets&rt=nc&LH_ItemCondition=1000
Ebay listings currently (5-23-2012) show over 200 Android 4.0 tablets selling for $83-$107:
http://www.ebay.com/sch/iPads-Tablets-eBook-Readers-/171485/i.html?Operating%2520System=Android%25204%252Ex%2520Ice%2520Cream%2520Sandwich&LH_BIN=1&LH_ItemCondition=1000&Type=Tablet&_pppn=r1&_dmpt=US_Tablets&Internet%2520Connectivity=Wi%252DFi%7CWi%252DFi%2520%252B%25203G%7CWi%252DFi%2520%252B%25204G
What do you think?
This question is probably off topic.
But here are my thoughts:
If you want to develop the truth is you should try to have access (be it ownership or just access) to any devices that you intend to support with your applications. This is certainly not possible for everyone, but there are ways that you can maximize your value while building a device collection by aiming to hit the largest swath of different types of devices.
You have no obligation to be locked into a contract (or even any service) for a telephone. If you go to a carrier store you can purchase a device at full price and leave the store without activating any service on it (In general, subject to change at the whim of whatever store you are in =) If you intend to by a phone for full price expect the range of prices to be $380-$600. From what I can tell about your situation I would think that a phone is best route for you to start off with. In addition to being able to buy from a carrier, Google also from time to time directly sells a "Nexus" device which is released with stock(ish) android OS (no skins etc..) The current device in this line is the Galaxy Nexus which is being sold directly through the (former) Android Market That is a good choice because it is more or less the latest generation of hardware, and the newest OS (these Nexus devices tend to receive their OS updates in a more timely manner than carrier tied devices). In fact the first Nexus device was marketed heavily toward developers, this line of phones was meant to be used for development.
Is a tablet PC the type of device I should get?
If you intend to create applications for tablets then yes. If you intend to develop phone applications you should really be testing on a phone.
Should I get a device with Android 2.3 or 4.0?
Ideally both on different devices. If you must pick one, then it depends on your budget 4.0 is only on the newest devices right now so they are likely going to be more expensive than some of the devices you can find with 2.3. (although $399 for the Galaxy Nexus might be hard to beat in price, even for some of the 2.3 devices for sale)
In addition to wifi, what else should I look for?
On phones that can be readily purchased in the US your choices are basically boiled down to only a few things. Screen Size (anywhere from tiny to pretty damn big for a phone), HardwareKeyboard(in a few different shapes and sizes), Camera (if you care about it). In general most of the other features are fairly standard (i.e. bluetooth, gps, accelerometer, etc..) The other thing to consider would be "Oomph", despite being called phones what people are carrying around today are small computers. They have CPU, GPU and RAM just the same as PCs. Battery life is the last major divisor, there is a fairly wide range of battery life expectancy (hint, massive screen and multiple cores need lots of battery). However since you don't seem to want to use this device as your phone some of this stuff may not matter as much to you.
Are there any brands/models I should avoid?
Anything in the list of your first ebay link. In general (in the US) the major phone manufactures are: HTC, Motorola, Sony, LG, Samsung. There are many other devices out there that are nice as well. But these guys are generally the ones consistently pumping out the most used phones. Idealy you should aim for a spread of device's made by different manufacturers. The custom skins and addons they like to use in their versions of the OS tend to like extra testing. Having one of each will help you help the largest chunk of your users.
Is there any point in buying used?
There are good deals to be had on nice but slightly dated phones. If you don't know what you are looking at I'd stay away though, it could be easy to get ripped off.
Well it does not seem like you need a very expensive device. Guess you could buy Google's and Samsung's old Nexus S which is not the newest around but still has android 4 because it's Google's "offcial" or something. And since it rolled out with 2.3, you should be able to download the old version somewhere and flash it. That phone should be cheaper, especially second-hand.
I mean since there are so many devices with different screen sizes and all you can't really find one that will reveal what other device-users will think of your apps.
Carrier options should be a non-issue for development. You can buy any Android phone for any
carrier including unlocked, and that should not affect your development flexibility.
Get the lowest version you can have. 2.3 and 2.2 have the most market shares right now. There is no reason to get 4.0, unless you are developing specifically for Android OS 4.0 and above (meaning that you actually use API that only exist on 4.0 and above and not lower).
Your development platform also affects what phone model to use. If you use Windows, get a phone that comes with Windows drivers. It'd be a bummer to buy one only to find out that Windows can't detect it. Mac and Linux do not need drivers, as they are both UNIX based.
You should get a phone. If you don't want to use it you don't have to... but most likely the majority of your users will be on a device. If you want a device that is always up to date and is great for testing I recommend the Galaxy Nexus.
Since most Android phone users are not on the most up to date os version, I prefer testing on a phone with 2.3
If you temporarily need a phone to test your app, Sony Mobile developer program has a phone loaner program in US and Canada where you can borrow a device for up to 30 days.
http://developer.sonymobile.com/wportal/devworld/phones/borrow-a-phone
/Magnus E
Sony Mobile
Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 9 years ago.
Improve this question
I'm going to be writing my Android app soon, and thinking about buying a solid android device to develop on and play with. My question is for you guys, assuming you wanted to develop and test your app on an Adroid device, what would you recommend?
Samsung Galaxy looks promising, but I know choosing a good dev phone is one of those things you definitely need to choose wisely on. The Android device landscape seems to be broadening every day, so it'd be good to hear about the latest devices for developing. Perhaps I could just get a cheap clunker and it won't make a difference?
I'm sort of a nerd, so if the device is pricey, but has some cool features that others don't, then I'd still consider it. However, I don't want to waste too much money on features that I may not need (or that may distract me ;)), like HD video, although it looks like a lot of phones have HD support anyways.
Tablets seem a bit much, but I'm still open to it... if you're loving your Android tablet for developing, let me know about it. I could see some benefits like being sure my app will run on Android tablets and different resolutions.
Any caveats I should be worried about? I have a buddy who was developing Android when it first came out and I remember him complaining that he could never update his device, since the updates were always restricted for some reason. Is that still the case?
If you can speak to any of these points, I'd be happy to hear them.
Also concerned about using the location based stuff, like gps, will I need a contract?
I'll be developing on my Macbook Air OSX, if you have any additional input as far as devices go, and using a Mac, that would be great too. Thanks!
You should develop against the Android 4 platform, but during the course of development, you will figure out which APIs you are using. The level of APIs you are using determines which minimal OS you will support. There's 2 manifest settings in play here: target API (15 aka ICS), and minimal API (depends on what you use).
To get a good mix of API support and broad coverage in the consumer market (using the dashboard that #Basic pointed to), I tend to point people at API level 8, which is Android 2.2 (aka Froyo). You'll hit roughly 93% of people out there and you'll get the APIs you need for most development. If you know you need newer APIs, by all means use them. The world will catch up, it may just take some time.
I typically point people to get a Nexus phone or XOOM tablet. You need the reference design so you can stay on top of the official releases from Google as quickly as they come out. The AOSP builds are important to follow, but they aren't the only game in town. After that, if you want to be serious about it, you should pick up one phone each from the major OEMs so you can deal with the differences in the customized platforms (Sense, Blur, Touchwiz, etc.)
Disclaimer: I work for Motorola. MOTODEV Studio is my teams' product (thanks Basic!)
Samsung Galaxy Nexus, Good because it is a phone that is supported directly from Google.
Samsung Galaxy S2, it just has all the features you need
One of those two for spec heavy phones, but you also need a cheaper not so good phone so you can get both ends of the device range, so get something like a IDEOS as well.
All you need to remember is that there is never going to be just one type of device being used, there are hundreds, if you have a phone thats close to the bottom and your app works almost as well as a phone as the top it will be good, because you will be able to catch the whole range of customers.
Also for location based things all you need to do is put permissions in the android manifest and it should be good, users are notified about these when they download the application
Mac can be good for development, its supported by Android as well so I dont think you will have a problem, when I went to the Android devs lab it seemed to be that all the Google guys had macs anyway
It depends what you want to support. There's a whole spectrum from the HTC Wildfire up to Galaxy S2 / HTC One.
What versions of Android are you going to target? Some people are still on 1.5!
Regarding updates: Google releases updates at their own speed. The problem is that before that update reaches a given device, the carrier usually customises it to include branding, mandatory aspps, etc... Which can take an indeterminate length of time or may not happen at all. This was one huge benefit of the Nexus One - it wasn't branded so updates were available as soon as they were released.
Regarding Development: eclipse is available for almost every platform and is a great starting place. MOTODev Studio is based on eclipse and is excellent as it comes with templates, pre-configured emulators, etc...
Location Services: Android uses 3 approaches to get your location: GPS (Works poorly indoors, can be slow to get a fix, quite a battery hog), Wifi (Google have mapped the location of a lot of wifi networks - thus if it can see a known wifi network, it knows roughly where it is) and finally network (It knows the location of the cell tower(s) your phone is talking to).
None of the above needs a contract but Wifi requires an internet connection (to perform the lookup) and Network obviously requires your phone to be in range of and allowed to talk to a cell tower - which usually just means getting a sim for a network which covers your area.
Note that when getting location on the device, you can opt to have coarse of fine-grained location information. Coarse doesn't fire up the GPS (They also require different permissions)
See this page for more info about location
Personally, I would recommend the Toshiba Thrive 10.1 Inch Android Tablet.
It has:
1 GHz Nvidia Tegra
1 GB RAM
Full Size USB Port
Full Size SD Card Slot
Full Size HDMI Port
Mini USB Port
Can be upgraded from Android 3.1 to 3.2 or 4.0
And it usually only costs about $200-250 depending on how much memory you want.
I have an android app with more than 500,000 users. I want to try to port it to WinPhone7, but I haven't any smartphone with WinPhone7. Is a real device needed to publish an app on WinPhone? Is there some developper phone?
First of all, I will say that for some scenarios, there is no real substitute for have a physical device to test against. Having said that, I would suggest that 99% of what most apps will do can be developed and test perfectly well on the emulator that comes with the developer tools.
The advantage of the emulator is that you can write and test without shelling out for the hardware and then signing up to create.msdn.com to get it (officially) unlocked, but once you are ready to deploy to the marketplace you will need to sign up anyway.
In your case, I'd say the main word in your question is "try". You don't seem confident in being able to port to the platform so the emulator route seems like the best starting point.
Your will find a Windows Phone 7 emulator in the Windows Phone SDK. You can download it for free on create.msdn.com.
There is an Android to Windows Phone API mapping tool and Windows Phone 7 Guide for Android Application Developers white paper as described on the Windows Phone Developer Blog that you should find very useful.
For getting a development device, you should reach out to Brandon Watson or your local Microsoft WP7 dev rep.
Simple answer - no, you don't. There are plenty of applications out there that were published without being tested on an actual device. Whether it's a good idea or not - that is the main question here. Depending on your application type and its behavior, you might actually need a device.
Also, another problem is the fact that the resources used by the emulator are different from the resources used by the actual device. That being said, if your application runs just fine in the emulator, it will not necessarily run the same way on a device.
You can use the WP7 emulator to test your application. But if you want to deploy it on a real phone, you will have to unlock it though the App Hub portal. That will cost you 100 dollar/year though.
As you and others have rightly pointed out, you can start porting your application using the emulator. There are differences in the emulator and real devices. In particular, to answer your question, emulator does not take pictures.
If your Android app really has half a million users, MS will happily give you a developer device (nearly) for free.
Contact #BrandonWatson or #FrankPR on Twitter.
From my experience I can tell you, that the emulator works very well. But once in a while you will stumble about a problem that you don't understand why it happens. Then you try it on the phone and it works... So... The answer is yes...not!
We want to place a device in a store that operates as sort of a kiosk device. As in customers walk up to it and start interacting with our custom app. The app could be developed for Android or the iPad, so I'm trying to decide which one to use and would like comments on the following ideas:
Is it possible for Android or iOS to access services over the USB port? This would enable us to disable the network.
Is one particularly better for 24 hour always on?
I like the iPad as I think its supply will be more constant as we move forward and need to replace devices due to ones gone bad. Also, our app will probably work on future generations of the iPad. With Android, I'm not sure there will be that sort of consistency from the tablet vendors.
Kiosk mode? I think with the iPad by putting it in a kiosk case that removes access to the home button and turning on the restrictions we'll get what we want. What about Android? I'd rather not have to get into rooting devices and replacing their firmware.
Remote control? Any way to remotely control iOS or Android in a standard means? Our app will be a client to a master server which will obviously be able to control the app somewhat (when used purely as a display device to a customer, controlled from behind the counter).
My feeling is that neither Android tablets or the iPad is best suited for this. Are there other options?
I will try to answer your points, but know that I am probably biased towards Android, because that is where my experience lies.
With Android 3.1+, at least with the Xoom, you have full USB host capabilities. Things like USB flash/hard drives, keyboards, mice, even digital cameras, all work. If you need custom interop with a USB device, you could go as far as to write a driver for it.
24 hour always on is not good for any device with a battery, but neither is better in this situation.
While android apps are forward-compatible, bad programming practices and/or deviations from "vanilla" Android software and hardware CAN break forward compatibility. That being said, if you grab a Google Experience device like the Xoom, you won't meet as many surprises.
In Android 3.0, the navigation bar is built-in at a low level, and it is not possible for apps to remove it. Therefore, it is trivial for anyone to break a "software nanny."
I know that it is possible to control android devices remotely, but without knowing your specific needs, I can't really offer more information than that.
Good luck!
iPad NOOO believe me I am a convert to Apple for my home and business but when we went to launch kiosk the iPad FAILED Big Time.. Here are a couple of little (Big) issues we ran into.
If the device reboots you cannot auto launch you iPad app since Apple does not allow that.
There is a serious memory leak in the iPads browser. We were running javascript / CSS3 and it cratered intermittently. I literally spent 2 hours "today" on the phone with Apple getting the MAJOR run around. I finally said let me speak to an Enterprise Sales Manager as my project could mean thousands of iPads and I got NO WHERE. One Apple employee even told me they don't have enterprise sales managers.
If those weren't enough even though we are just in the proof of concept phase, we are already getting request for other options. These other options are going to require access to the OS which Apple yea right. We are moving to Android immediately.
Sorry Apple I love you but you loose here.
If your using an Ipad you should consider if it can support the power for the USB thing. Watch this Using Powered USB Port
Your idea about putting the tablet behind another piece of glass/plastic is neat. To then deal with remote controlling, you might consider doing some Bluetooth programming.
My mobile development has been primarily with iOS, so I am biased toward that SDK. I will mention that the data/sync/charge port for iOS has (I believe) never changed. Your Gen 1 iPhone sync cord works on your iPhone 4... and your iPad or iPad 2. So, in terms of third party hardware, you may see more consistency with Apple.
I haven't found a good answer regarding whether it is easier to do Bluetooth programming for iOS or Android, but I think to stay cost-effective, you might see which one is more open to third-party devices. Here is an SO post/answer about iOS and third-party Bluetooth devices; I've not found anything on Android regarding third-party Bluetooth remotes, but considering a lot of hardware running Android is third-party, your chances from a naive perspective seem pretty good. Here's the Android Bluetooth API.
Buying an iOS or Android handheld to remote control an iOS or Android tablet does seem a bit steep, but then again, maybe not. Cost also depends on your ratio of remotes to tablets. 1:1? 1:N? N:1? N:M?
The lowest end iPod goes for $229 as of May 20, 2011. Android does have more variety in terms of hardware. You may be able to get a cheap Android phone with no service plan to act as a Bluetooth remote for an Android tab.
I have provided a solution for the kiosk mode using iPad here Lock-down iPhone/iPod/iPad so it can only run one app
I am afraid that I don't really know for Android if the same thing is possible.
To address the issue of crashing applications you can use an exception trampoline (see discussion here https://blog.compeople.eu/apps/?p=275) to catch the crash and reboot your app.
If the entire device is restarted however then other apps that are on the device can be started and will subsequently be locked in.
To answer your other points:
You can use a configuration profile to control network access. Force it to use a VPN or Proxy that only allows your custom app with embedded credentials to use. That way other network access can be prevented.
Your concern over future compatibility is spot on. The Android marketplace is so fragmented then maintaining a fleet is difficult.
If you have an app that is behaving as a server and is locked in then remote control is possible.
We manufacture tablet kiosks that support both android and iPad devices. In fact we are the only iPad kiosk that has achieved apple approval.
Generally speaking i think you will have an easier time with an iPad as the software and hardware will remain more consistent over time. Which is important if you have to change out a fault unit or deploy more kiosks 6 months or year from now when the original device is no longer manufactured.
It's difficult to tell what is being asked here. This question is ambiguous, vague, incomplete, overly broad, or rhetorical and cannot be reasonably answered in its current form. For help clarifying this question so that it can be reopened, visit the help center.
Closed 10 years ago.
As the title says, how "good" is the Android emulator?
I don't have an Android phone but would it be possible to develop purely using the emulator as the testing environment?
My initial questions about it:
Is the emulator as responsive as a normal android phone (of course this depends on the exact phone but in general)?
Is there a way to control the memory on the emulator?
Has anyone had issues that have been flagged on the phone but not the emulator? or vice versa?
What's your normal testing procedure to test on both throughout, or emulator recursively and phone at the end?
Any feedback would be appreciated.
Thanks,
Q: Is the emulator as responsive as a normal android phone (of course this depends on the exact phone but in general)?
A: Absolutely not! The worst example is the 3D, which is, afaik, only software emulation.
Q: Is there a way to control the memory on the emulator?
A: Please specify, the size of the SD-Card is configurable, the RAM/VM-Memory... I don't know.
Q: Has anyone had issues that have been flagged on the phone but not the emulator? or vice versa?
A: Each device has his own customization. On the emulator you can only change the screen resolution, but caution: the size of your Emulator screen depends on the pixel density you set up. So what might look good on your big emulator could be too tiny on a real device. And don't forget, a finger is much bigger and not as accurate as a mouse cursor.
Q: What's your normal testing procedure to test on both throughout, or emulator recursively and phone at the end?
A: Functionality is tested during the implementation mainly on a (fast) device, like a Nexus One. The real test after implementation will be on as many devices as possible. The UI, too (most differences there)
Since I have a device which is has the latest SDK, I prefer testing on the device. Thats not necessary for programming an app, but please: don't publish an app without at least one test round on a real device :)
Yes, the emulator is very fast.
Yes, you can set the RAM size of the target you wish to emulate.
I'm sure someone has, but I haven't.
I usually test on the emulator during development, then on my G1 when I'm ready for longer term testing. After that I test on the emulator when checking compatibility with 1.5, 2.0, and 2.1 targets.
See here for what options targets support. Note there is a GUI tool for setting these up as well.
To answer the general question about how "good" the emulator is you need to look at what the emulator is.
The emulator is a piece of software (qemu) that is able to run arm instructions on a developer machine using emulation. This means that the emulator will run an android platform similar to what is running on a physical device. This setup is better than using a simulator where the platform is compiled for the development machine and not for the target hardware. There is however a number of limitations such as
peripheral support - a mobile platform includes a number of companion chips such as bluetooth, wlan, gps, radio access, graphics acceleration etc. All of these are interacting with the main CPU in ways that are not predictable and hard to simulate on the emulator. Some are supported in a limited way such as internet access while others are not supported at all such as bluetooth.
performance - the emulator is not a real processor and may not use physical accelerators available on an actual device. It also uses the host machine hardware for a lot of things meaning that memory access and similar items behave differently compared with testing on an actual device.
The general rule is that if it is possible try to do as much testing as possible on a device to avoid surprises in the end. The emulator is good for things such as testing general functionality, initial testing to avoid obvious bugs and checking that layouts and UI look decent at different screen sizes. When it comes to any performance critical testing or behaviour when interacting with the outside world, such as network interaction, the emulator is not recommended. It may tell you that your application has serious issues but do not trust "good enough" on the emulator to be good enough for the real world.
I'm going to answer #3 only, because the other answers have been covered already.
There's definitely some issues that are only present on the emulator and not on the device itself. One big one is that airport mode simply doesn't work on the emulator; internet still works despite turning airport mode on.
Also, if you're just using a stock emulator, you can't access the market, which means that any way which your app may try to interact with it can't be tested via emulator.
I know there are some other smaller issues which I've run into between an actual device and the emulator. There are enough things such that I highly recommended you don't develop solely on an emulator.
With emulator only your testing will have more mock testing to cover:
-touch
-sensor
There frameworks to plugin to simulate sensor/device movement, camera by using your webcam, etc.
You need memory to use emulatord like AVD emulator, emulator is good for testing but sometimes it can crash without reason.
Yes you can change this settings many times.
Some time ago I cannot emulate accelometer, but right now I can.
I prefer to test my app on real device. I prefer testing my app on real devices (if my frends cannot creash it and JUnit tests works then it shouldn't be a problem), but I test different screen size on emulators.