[my3681]

I'm not knocking this or any particular xxxx-to-mobile platform, but as a native iOS and android developer, I would highly recommend just learning the native languages and frameworks.

First, you have the absolute most control you will ever have over your application. This may not be true when a particular tool becomes popular, say at the release of a new tool, but this definitely shows up as time progresses. Who knows what tools will keep up with the native frameworks terms of functionality?

Another reason that is more qualitative, is that you come into contact different design styles, code styles and paradigms. Objective-C, for instance, is a tricky beast sometimes, but it is an interesting language with a lot to learn both good and bad. The new things you learn will most definitely carry over into your web programming!

The third reason to code natively, and I cannot stress this enough, is third-party library support. Maven and cocoa pods may dependency management much better on the mobile platforms, and using a third-party tool will most likely render all those advancements useless. There are some amazing libraries for both platforms that you definitely don't want to miss. They were reduce the amount of time you spend coding and improve your baseline quality.

[Touche]

> First, you have the absolute most control you will ever have over your application.

You have 0 control over distribution though, which is pretty important to me.

> There are some amazing libraries for both platforms that you definitely don't want to miss.

I don't doubt that the platforms have good 3rd party support, but it's absolutely dwarfed by that of the web community, and that will only grow larger over time.

[dmur]

> You have 0 control over distribution though, which is pretty important to me.

Simply having to wait a week to get approved for the iOS App Store (and less on Android) doesn't constitute 0 control over distribution. You can still release the app to the store when you want, if you plan ahead, and pull it whenever you like. Less control, yes, but not zero.

Also, how is this different from writing a non-native app? The same rules apply to distributing web apps if you want them to be in the store. True that you can release it as a website whenever you want, but that's not really the same thing.

[Touche]

> Simply having to wait a week to get approved for the iOS App Store (and less on Android) doesn't constitute 0 control over distribution. You can still release the app to the store when you want, if you plan ahead, and pull it whenever you like. Less control, yes, but not zero.

You don't have control whether your app will be accepted into the store (like the recent bitcoin apps), you don't have control whether the content in your app will be allowed in the store (such as the many Comixology comics that where banned from being sold).

> Also, how is this different from writing a non-native app? The same rules apply to distributing web apps if you want them to be in the store.

But I don't have to put my web apps in the store at all, I just tell my users to visit a URL.

[rimantas]

  > I don't doubt that the platforms have good 3rd party support, but it's absolutely dwarfed by
  > that of the web community, and that will only grow larger over time.

You know why it is funny? The things that native offers without even third party code is leaps and bounds beyond what web community offers for mobile and will stay this way for a long time, if not forever. So far community has been reinventing the same wheel a thousand times over and there is no sign of progress. Guys, look, another MVC framework, how cool is that!

[marknutter]

So if you want a web app, an android app, and an iPhone app, you have to code it up 3 different times in three different languages. The reason why the web exploded in popularity as a development target is precisely because it made it easier to target a larger number of customers. Imagine if Facebook had started out as a Windows XP / Linux / OSX native application. It never would have even got off the ground.

My take on this is that if you absolutely need the performance, go native and pick the platform you'll get the most customers on. If you're app is a glorified CRUD wrapper, though, HTML5 will probably suit your needs and you'll only have to code it up once.

[tluyben2]

I don't buy it. Enterprises definitely buy into that idea and they do have their employees work with these crippled apps because they have no other choice. But the experience for this 'CRUD' app is definitely not good enough. It's by forcing it that it gets used. It depends on how intensively it will be used in reality, but I definitely know quite a few companies who went from BYOD and an HTML5 Cordova app which was very intensely used to just buying Android devices for the employees and making a native app to boost productivity.

If you have to enter / search / whatever information in an app which is horrible to use (a BIT of lag can ruin your day; you tap/swipe; because of the lag it just responds a bit late, the keyboard pops for the wrong field or you submit accidentally; plop seconds lost and so is your good mood) all day, you're not going to want to work with crap. It's not 'good enough' even though management might think so.

[marknutter]

You don't have to buy it, it's an undeniable reality. Yes, if you're an enterprise and you can afford to write native apps for all platforms, then that's the better choice. If you're a small startup or a single developer, writing the app 3 times is just not cost effective.

[tluyben2]

Well then there are still a few options; 1) it's internal and you can force-feed it to your employees, then it might be ok 2) it's external and you're trying to get a lot of people to work with it outside your company.

In both cases I still think you are better/easier/happier/cheaper off building it once and in case 1) giving your employees tablets/phones with the target OS and in case 2) just waiting till you have enough critical mass to warrant writing a version for the next OS.

Not to mention that, in most cases we have made HTML5 'hybrid' apps, we noticed that it takes more time making it uniform and smooth across the plethora of (especially Android) devices than just writing 2 native versions. And the HTML5 version will just never work well on a large amount of 'Alibaba $50' Android devices which surprisingly many people have. Resulting in TONS of bad reviews (if public) and/or very frustrated people.

[dmur]

Depends on whether or not quality of the product is a factor in your cost calculations. It should be, and for many startups and solo developers, it absolutely is.

[rimantas]

Most likely you never tried to do what you preach. You will end writing your app only a little bit faster (if you are lucky) and then three times more on debugging.

[marknutter]

I have. I built an HTML5 hybrid mobile app for Android and iOS. It was about 90% JS/HTML/CSS and 10% native code. Worked very well for our purposes. You can judge for yourself, it's called Kona and in both major app stores.

[tluyben2]

Eh yeah. Exactly what I said then reading the Android reviews then? Works on some devices, terrible on most; getting that JS/HTML/CSS to work well on 'most Android devices' is infinitely more difficult than doing it native. Is that worth it? I still want to bet I could've written both apps native in a shorter time with much better results. You can take me up on that any time.

[auvrw]

> 90% JS/HTML/CSS and 10% native code.

this was going to be my comment on the original comment: it's not like you have to do only HTML5 or only native, or so i'm given to understand... i'd be interested to hear more about the hybrid approach if anyone has links/blog posts.

[archagon]

Pick a platform, port later if needed. It's what most indie app developers do.

[Iftheshoefits]

No, the web exploded in popularity as a development target because the labor pool is so diluted that it's easy, quick, and cheap to hire (and fire) any random "webdev" who is good enough for their purposes. And that's why, unless I get desperate, I wouldn't put PHP on my resume and would only list non-webapp projects for other languages I've used professionally but which are also used for webapps (e.g. python)--I don't want to put myself into that (cheap) labor market.

[CmonDev]

You are forgetting about the modern contender - Xamarin. It's a sweet spot between HTML5 and native: modern high-level language and a very decent performance.

[skrowl]

With Xamarian being acquired by Microsoft, I'm really hoping to NOT have to learn native android / ios / windows phone development. I don't have time for that, and I'm already an expert in C#.

[city41]

I regret investing in MonoTouch and wish I had instead just learned native iOS dev from the getgo. You need to at least be comfortable with native iOS dev to be effective with MonoTouch anyway. I have a MonoTouch written app that I can't do anything with unless I pay for a new version of MonoTouch as it's no longer compatible with current iOS releases. The constant song and dance between releases of MonoTouch and iOS updates gets frustrating after a while.

[tluyben2]

I always recommend (and when it's about employees demand) people to first learn everything native in Objective-C/Java(/C# if you need wp8 as well) before trying to unify things with C#/Xamarin, C++ or JS. That way you have a solid understanding what is possible and you don't lock yourself in. That said; Xamarin rocks and I hope MS doesn't kill it.

[Varcht]

As a C# developer I am well into a project using Xamarin. I have not run into any road blocks and I am very happy with the performance on both iOs and Android. I currently have 80-90% shared code utilizing portable libraries and MvvmCross. Only real learning curve has been at the UI level but they are both rather WPF'ish especially Android. I highly recommend this path for a C# dev.

[jawngee]

The language isn't the issue, the frameworks and libraries are. You're going to have to spend your valuable expert time sorting those out too. That's the bulk of it.

If your truly an expert at c# then learning objective-c isn't much of a challenge. At least learning the amount you need to build apps.

[yulaow]

The advantage of using a single language is that, for each single app in all the three store, you have to change a single codebase and not to make 3 different changes in 3 different codebases just because each of them require its own language (and then its own specific libraries). So you have surely to spend the time to sort the libraries, but it is a one-time problem. After it, you can just use them without problem in all your projects and you really can focus on your idea

Put all the code in a good ui/codebehind separation framework like mvvm, and you are happier than ever.

[RyanZAG]

It's a nice dream. Pity reality is so very harsh on nice dreams.

[bmj]

See my comment above: I am under no delusions that cross-platform frameworks are truly cross-platform on the UI level. But, for our app, the UI layer is roughly 50% of the app. What about the core of the business logic? The code in a cross-platform framework is truly cross-platform. How do I approach that doing full native development for Android and iOS? Is there a way to create a cross-platform library?

[jawngee]

It's called C++.

And I can tell you, from my own experiences, your backend isn't going to be as cross platform as you dream of it being because everything from networking on down is done differently on Android and iOS. Which isn't to say that you can't front-end all of that with a shim in C# or C++ or whatevs, but on mobile, the bulk of the hard stuff is UI anyways and there is no cross-platform panacea as of this writing.

[bmj]

As someone who has used several cross-platform frameworks (Titanium, Cordova/PhoneGap, investigating Xamarin), I have a question about going "full native": how do you handle shared core code? Our app has an engine that drives the business logic, and the cross-platform frameworks have been useful because we don't have to deal with modifying the core codebase for different platforms [0]. Is it possible to create platform-agnostic libraries that can be shared?

[0] Despite the many claims of "cross-platform UIs," we have many cases in our UI layer of doing different things on different platforms.

[tluyben2]

Xamarin is great for sharing code. You just rewrite the GUI layer per platform. It's great, cannot really say it any differently.

[CmonDev]

And that GUI layer could be just Views in the end, since it seems you can even re-use ViewModels/Controllers.

[mattgreenrocks]

This is the exact use case for Hexagonal/Clean Architecture: http://blog.groupbuddies.com/posts/20-clean-architecture

It is contentious on HN, but I implore you to draw your own conclusions.

[untog]

If have read before that C++ can be used to share logic. An interesting development with iOS7 is that it has a JavaScript runtime of some sort (I don't know the details) - perhaps it will be possible to use JS for cross-platform code.

[notjosh]

You're thinking of JavaScriptCore. It allows Objective-C objects to be used inside of JavaScript (and vice versa).

It's actually a pretty cool technology. I gave a talk about it this month at Berlin CocoaHeads, and I'm gonna put some blog posts together about it soon!

[mattgreenrocks]

I presume the JSCore runtime bridges objects back and forth? Very cool!

[wil421]

>I'm not knocking this or any particular xxxx-to-mobile platform, but as a native iOS and android developer, I would highly recommend just learning the native languages and frameworks.

First off I agree that native wins over some bundled HTML5 app anyday of the week.

But for many reasons companies still choose to use phone gap or another framework to compile HTML into a native app. The first and foremost reason is that a lot of organizations have plenty of web devs available with skills to build web apps.

Managers would rather leverage the skills they already possess. Believe me I do web dev as part of my job and I still try to make the case that Native is better for mobile. In the end it is cheaper to use developers they already have with tools the devs a familiar with (HTML/JS) than to hire several new devs to code against different mobile OS's.

Right or wrong Managers and Organizations like cheaper and faster.