[archgrove]

I totally agree with some of the points: iOS should really receive the same garbage collector that OS X has had for years, and the provisioning certificate nonsense is, well, nonsense. iOS really does need a side loading mechanism. That you need a Mac to develop on is, I suppose, a negative - you can get going with Android on almost anything.

However, I can't say I've ever had any problem with Apple's documentation: It's clear, well written, generally entirely correct. I must confess, I've never spent "weeks devising and performing increasingly peculiar experiments to figure out how to get iOS to do what [I] want", any more than on any platform. If he's complaining that iOS has private APIs then, well, I'm quite sure Android does as well - private just means "not guaranteed to exist in the same form on an upcoming release". If he's claiming that Android's "openness" allows him to see deep inside the OS to make design decisions, rather than relying on the documentation, then I'd suggest that's a mad development strategy (unless one likes rewriting when new OS releases come out).

The point about the simulator seems to be that Android's is so bad, you have to use the phone. I can't really see that as a plus, as one could do exactly the same thing on iPhone, except that iOS has a working simulation environment for when you want it.

The remaining points, about the initial user experience and development environment are entirely subjective, so one can't really comment either way. His point that developing for Android seems to be "easier" than iPhone runs contrary to my experiences, but what one man finds easy, another might find hard.

[rgbrgb]

Agreed. I had the exact opposite reaction when I started doing some Android dev after coming from iPhone.

I find Xcode to be at least 80,000 times better than Eclipse (memory usage, UI, interface builder, speed, general bugginess).

I also REALLY like Apple's docs and the ease of integrating C code (no NDK!) when you need to do something like real-time audio.

My only really big complaint is the certificate signing process which can be a real pain.

I mean, I can totally see why Android might feel better or more familiar to a Java programmer though.

[BenoitEssiambre]

I have a question. I asked this to a few mac users and haven't received a good answer.

How the hell do you get xcode (or other programs, but xcode is particularly bad) not to end up being a big piles of small windows you can't access effectively because they don't have a dock icon? The only way I found, was to long-click on the xcode dock icon which after a while splatters small versions of the windows everywhere, then scan these tiled windows until I find the right one and click on it. I have to do that atrociously long multi-step process every damn time I wan't to take a glance at another window! This, for example, makes the internal iOS documentation useless to me. At least I can use the web documentation to get the browser dock icon but when I don't have an internet connection I'm out of luck.

While I'm at it, is there any way, when using 'Spaces' to do a desktop change in one click? I'm mainly a Linux user and I'm used to having multiple desktops. OSX also has this functionality, and the 'Spaces' icon actually has four little square on it that represents the four desktops it's controlling. However when I click on one of the small squares, instead of going to the right desktop like it does in Gnome/Linux (and has been doing right since about 1995), it goes into an animation where the four desktop are displayed tiled full screen and I have to pick one. That is a two step process with an animation in between for something that should clearly be instantaneous. Is there an alternative to this. Both these things are driving me insane!

Like in the article, I bought a mac just to do iOS development and up to now my experience on osx felt like using a broken out of date gnome desktop with serious usability issues.

[metachor]

To use XCode in a single window, switch it to the All-in-One layout in the preferences:

http://iphonedevelopment.blogspot.com/2009/03/xcode-single-w...

Also, in any OS X application with multiple windows you can use Command-` (i.e. Command + backtick key) to switch between windows of the active application.

[allenp]

I'll add this because it has happened to me - sometimes you're in all in one mode and you still don't see your code, just the file name that you selected at the top of the window - this is because there is a horizontal divider that is pulled all the way to the bottom of the window. Look for 1 little dot at the bottom of the window in the center, if you drag that up you'll be able to see your code again.

[BenoitEssiambre]

All-in-One is kinda nice. It seems it should be the default. It still doesn't make the documentation window of any use though. I guess it has a little less chance of being hidden under a pile of other windows now. Command+backtick doesn't seem to do anything on my mac.

[losvedir]

Strange that Cmd-backtick doesn't work for you. Try this: Open up System Preferences, select "Keyboard", change to the "Keyboard Shortcuts" tab, and highlight "Keyboard & Text Input" on the left. Is the "Move focus to next window in application" box checked? What's its shortcut?

Also, long-clicking on the dock icon just activates Expose for Application Windows. I have it set-up to activate upon mousing to the top right corner and really like it. You don't have to wait for the delay of clicking and holding on the dock, though you'll still have to scan for the correct window.

[CountSessine]

I wish I could vote you up more than once. Not using all-in-one layout in XCode is Considered Harmful. This is even more important than binding Open Quickly to CMD-O or some other quick key shortcut.

[radishmouse]

I've got a few tips that might make your experience a little better. Option 3 below will be particularly helpful for accessing the documentation more easily.

Option 1: Re-map your Expose´ keys. I mouse (tablet, actually) right handed, so I use the following shortcuts that I can access them quickly with non-mouse hand:

F1 - All Applications

F2 - All Windows

F3 - Desktop

F4 - Spaces

This allows me to use Expose´/Spaces via keyboard in tandem with the mouse. While technically two steps, it feels more like a single, coordinated step to me.

I use XCode in the Condensed (not All-in-One) layout, which results in lots of small windows. I hit F2 (or F1 if I've secluded XCode to a single Space), then either:

- mouse over the various windows and press the space bar to see a zoomed-in view of that window. Then, click the one you want to bring into focus.

- press an arrow key to highlight a window, press space to see a zoomed-in view, press arrow keys as necessary, and then either press F2 again (or click the left mouse button) to exit Expose´.

Option 2: Press Cmd + Shift + D. This will bring up the "Open Quickly" dialog box. Start typing the name of the file you need to open or bring into focus.

Option 3: Use shortcuts to go immediately to the definition of a class/method/protocol/etc., toggle between .h/.m files, or open the documentation to whatever's under the mouse.

- Press Cmd key then double click a class name (or method name or whatever). This immediately opens a window to the definition.

- Press Cmd + Option + Up Arrow to toggle between .h and .m files for a class

- Press Option then double click a class/method/etc. name to open the documentation in a floating window.

- Press Cmd + Option then double click a class/method/etc. name to open the documentation in the XCode documentation window.

Mac OS X is littered with these kinds of accelerated interface shortcuts. I wish I could point you to a good, consolidated guide; but I have yet to find one on the web. Several of them can be found in the opening chapters of Aaron Hillegass' Cocoa programming books.

[jamesg]

Don't know on the first one, but you can use ctrl + cursor keys to move between spaces. Fairly sure that's the default, but if not, it can be setup in preferences for Spaces.

[BenoitEssiambre]

It doesn't seem to be the default but I guess I can find out how to set it up. I still think you should get a mouse based one click interface though.

[jcoby]

Ctl+1 through 9 switches spaces directly. If you enable the spaces menu bar in system preferences, you can click on the icon and choose which space to switch to (in snow leopard at least).

[mxcl]

Grab Xcode 4. It's still beta and buggy but it's all one window and is pretty good IMO.

There's keyboard shortcuts for spaces. And CMD` to switch between windows in apps.

OS X isn't perfect, I switched from KDE 5 years ago. Eventually you start noticing all the little touches that mean it is—really—light years ahead of Linux desktop options.

It does have a lot of little "Apple" touches though that you'll just have to take with a LOL.

[mblakele]

Another tip: ctrl-shift-d brings up the "File > Open Quickly" dialog, where you can type the name of a file into a filtered list (eclipse also has something like this). So instead of finding the right window, just bring up whatever file you need via the keyboard. I find this faster than using the mouse or cmd-`.

Another shortcut I used frequently is cmd-shift-e. This toggles "View > Zoom Editor" so that I can see more code.

[raquo]

Well, there's Cmd + ~ hotkey for switching windows of active apps in 10.6, and it can be added using some small app in 10.5

[frio]

It existed off the bat in 10.5 :).

[GeneralMaximus]

Cmd+` has existed for almost as long as the MacOS itself - and I'm not just talking about MacOS X here. Interestingly, BeOS also used the same shortcut, maybe because a number of Be engineers initially worked at Apple and the original BeOS was Mac-only. You can also use Alt+` in GNOME with the same effect, though you might have to enable a setting in GConf first.

[elai]

There is also an expose option to show all windows for an application. Set it to a keyboard hot key.

[jamesg]

I'll make just one refutation to this: I am not a Java programmer :)

I had to learn Java specifically for this project. Python is my preferred hammer for most nails, but not an option on mobile. I've also been a professional C programmer before, wreaking havoc in the kernel. I've got opinions on Objective-C, but that's a subject that deserves a whole separate post.

[biafra]

Just a quick reminder: You don't have to use Eclipse to develop for Android. You can use any other Java IDE (I use the free version of IntelliJ IDEA) or no IDE at all. You can even use a combination of XCode and Maven to develop for Android.

[rimantas]

Garbage collection on OS X appeared in Leopard with Objective-C 2.0—later than iPhone, but before iPhone SDK. Older models are still around and even iPad has less memory than iPhone 4 so it is not unreasonable to treat it as resource too precious to be left ar the mercy of GC. On the other hand, manual management is not that complicated and does not add much overhead in programming except the initial phase of getting used to it.

[rbritton]

If they do ever introduce it, I sincerely hope it is completely optional. On my iPad app, I run with such tight memory constraints while dealing with large image files that it has to be freed up when I tell it to or I'm likely to get killed by the memory watchdog.

[chunkbot]

The existence of a garbage collector doesn't mean that the programmer has to play a totally hands-off role in the allocation of objects; as with any coding, there are usually a few bottlenecks that deserve special care. A system with a good GC will provide opportunities to tune; for example, to make hints to the GC about lifetime and locality and so on.

http://www.jwz.org/doc/gc.html

[protomyth]

I would expect it would be if they follow the pattern from OS X. I do wonder what the breakdown in new OS X apps between GC and retain/release.

[chunkbot]

I treat memory as a resource too precious to be left to the mercy of programmers. In any application, a good garbage collector can be more efficient than malloc/free or new/delete.

http://www.jwz.org/doc/gc.html

[rimantas]

How about compared to retain/release/autorelease?

[SeanLuke]

?? I admit the last time I really coded in Obj-C was for NeXTSTEP. But back then init ultimately called malloc, and release ultimately called free. In any case, yes, a good GC can be faster as it takes advantage of cache effects.

[chunkbot]

I upvoted you, but I was mostly just quoting the jwz article from 1998.

[chc]

Older iPhone models cannot run the current OS anyway.

And manual memory management is not a showstopper, but to say it's "not that complicated and does not add much overhead" is just wrong. It's conceptually simple, but the devil is in the details. I would bet that if you took a poll of all Cocoa programmers who have been working in the field since — well, pick any date you think qualifies as "out of the initial phase" — and ask them how many of their programs have done correct memory management without debugging, you will get an answer of 0.

[rimantas]

  Older iPhone models cannot run the current OS anyway.

This is true only for the oldest (1st gen). iPhone 3G and 3GS run 4.2 just fine. Sure, some of the most prominent features of iOS 4 are missing on 3G, but it still does not qualify as not running.

[tres]

I don't disagree with the premise that memory management requires more work; however, the tools and utilities that Apple now provides to hunt down memory issues makes it more tedious than difficult.

[mickdj]

Clarification: We don't use any private API's on iOS, James' comments were with respect to what we do with location and networking.

Apple's documentation is great for most visual elements, but CLLocation* in particular has quite flawed documentation.

[pkaler]

Where are the flaws in CoreLocation or the documentation? CLLocationManager is instantiated like any other NSObject. CLLocationManagerDelegate returns asynchronous results like any other protocol in Cocoa/CocoaTouch. CLLocation, CLHeading, CLRegion are about as close as you're going to get to get to C-style POD structs in Objective-C.

The documentation is all here and the API is about straight forward as it is going to get. http://developer.apple.com/library/ios/#documentation/CoreLo...

[mickdj]

I don't want to veer O/T, but that's not the hard stuff in CLLocation.

There's a lot that could be discussed, but as one example: optimizing for battery conservation requires knowing which radios are currently powered up.

iOS makes its own decision as to which of the 3 styles of location service to engage, based on the desiredAccuracy and distanceFilter values.

However the WiFi/GPS radio have different costs for runtime and warmup, so this API doesn't help when you are attempting to optimize for all 3 of: Accuracy, Timeliness, Battery conservation

Furthermore, the accessible battery percentage in UIDevice.batteryLevel is only reported to the nearest 5%, which is not granular enough to be of use in real-time server-based tweaking.

[pkaler]

There's a lot that could be discussed...

There isn't a whole lot to be discussed. iOS minimizes battery usage based on how close cell towers are, type of cell tower, altitude, Wifi spots database, state of GPS almanac, hardware/driver combination, etc like a any good OS/kernel (including Android) should.

The App sets CLLocationAccuracy to kCLLocationAccuracyBest, kCLLocationAccuracyNearestTenMeters, kCLLocationAccuracyHundredMeters, kCLLocationAccuracyKilometer, or kCLLocationAccuracyThreeKilometers and that's it.

The OS will almost always be able to minimize battery usage because it has way more information than the App could or should ever have. The App will never be able to optimize this setting because there will always be new hardware with new tradeoffs.

Your claim is that Apple's documentation does not describe the implications of the laws of physics for an arbitrary location that the user may reside at and does not reveal implementation details of the kernel that change every hardware/software revision.

The API could not be much simpler or much better documented.

Do you really think that Android's documentation is better? http://developer.android.com/guide/topics/location/obtaining... http://developer.android.com/reference/android/location/pack...

Because the Criteria class on Android is essentially the same as CLLocationAccuracy. http://developer.android.com/reference/android/location/Crit...

I really would like to know if you think that Apple's documentation is poor or if you are just trolling because the facts of the situation don't support your argument.

Which part of the documentation is "quite flawed" like you originally claim?

[jamesg]

So, I will say that I bundled documentation and "openness" into one box, which I probably shouldn't have. The connection is that absented of the "openness" we were really looking for, we sought documentation to describe what's going on. That said...

CoreLocation is a good abstraction. However, its primitives for specifying what trade-offs you are willing to make in acquiring a location are, well, primitive.

A couple of random examples:

- CoreLocation doesn't tell you the source of the location sample (GPS, WiFi, etc). It gives you an estimate of accuracy. Of note, it doesn't give you a measure of the accuracy of the accuracy. This is of import as we have seen examples where the data is off by a whole hemisphere -- I'm not kidding! I understand that exposing these details is kind of "ugly", but obscuring it is removing signals that we could use to figure out the reliability of the data, and what techniques we might be able to use to "clean" the data. I am willing to concede that CL is a good API for general use, but when you're building consumer products, that doesn't cut it. The guy in New York who was reported as being in Antarctica (again, seriously), doesn't really care that the iPhone doesn't provide us the tools to fix that, he just wants it to work (and he's no longer a user).

- Related to the first point, but separate: the implementation of the algorithm for seeking to the desired accuracy is a black box. This makes it really easy to use for basic stuff, but you really don't have any way of knowing the result, in milliwatts, of passing in a given value to that argument. There are ways to mitigate this (which we've had to explore), and we can experiment to learn what the drain is, approximately, but hiding that information has obstructed our development process. Consider also that CL doesn't allow me to specify how long I am willing to wait to get the location fix at the desired accuracy. It does not let me set a budget in milliwatts for a location fix. I realise that Android doesn't provide those exact abstractions, but the tools it does provide make it easier (by which I mean "possible") for me to build them myself.

I also don't think you've successfully made the argument that the OS will know better than us. It's a generic tool, which makes some assumptions. It will have made compromises that don't necessarily work for us. It will not perform optimally for every use case. Going back to something I said before, it seems to optimise for getting to the desired accuracy quickly. For background location tracking apps like ours, that is not a priority. Power is. Neither CoreLocation's abstraction nor documentation provide for this use-case.

[mickdj]

"The OS will almost always be able to minimize battery usage because it has way more information than the App could or should ever have."

This is only true for standalone apps which don't share location information between people.

The OS doesn't know that one person wants to turn on the GPS on another person's phone.

I'm actually pretty pro-Apple and anti-Android (James, OP is the Android developer and I develop exclusively on Obj-C).

[wankerrific]

I'm not sure I agree with the blog post then. The Android docs certainly do not detail this kind of information either.

In fact, I've found the docs for Android to be of the "broad but shallow" variety. That is they cover everything (thanks to Javadoc) but they don't necessarily provide usage guides on the various APIs.

My opinion on Android development vs. iPhone is:

a. GC is convenient

b. The developer docs on Android are not nearly as good as Apple's

c. XCode (while not perfect) is less buggy than Eclipse for basic development activities. Eclipse just gets in the way most of the time and can't keep up with my typing speed which is really frustrating.

d. I like ObjC's dynamism a lot more than Java. This is a personal preference.

e. Its nice to be able to peer into the OS source on Android as the definitive answer for API questions but I suppose if the docs were better you wouldn't need to do that.

f. The Android APIs are full of pattern inconsistencies with their implementation on the whole compared to UIKit which make them more difficult to learn.

g. UI Layout on Android is abysmal compared to using Interface Builder.

i. Again - this is more of a Java thing, but Android lacks the ability to do conditional compilation. You have to go through build system/scripting gymnastics to get what you would get from a simple #IF statement in C.

[elai]

There are some things that are in the private apis that make dealing with certain problems a 100 times easier. Sometimes you just want the gui to force a rotation to a certain orientation at some point of the app. Other public api's such as rotate top bar are more crufty, and don't work 100% of the time, or require fragile workaround where you do all the rotation manually yourself because the public apis are fragile, while the one-line simple private api would of done the trick.