Hacker News Reader

Comment by onion2k | Hacker News Reader

Unless you're making something that needs really good performance (3D gaming for example) building a hybrid app wins every time. Something based on Cordova (eg PhoneGap, Ionic) with Crosswalk can get you to about 90% of the performance of a native app, with about 1/2 of the development time because it's all built with web technology. 90% is easily good enough for most applications. Cordova wraps a uiWebView in some boilerplate to make a web app effectively native to give you the 'icon on the homescreen' advantages (selling the app, in-app purchases, access to most phone APIs) with all the benefits of building for the web (easy development, lots of developers, etc).

View on HN

Developing for native does not always take twice as long. I would write a desktop C++ app faster than attempting to write a web app and wrap it in desktop runtime capability.

If you're not used to writing desktops apps, yes it will take longer.

I recently needed to build an internal GUI app - after seeing how much time it will take me to get a HTML client/server set up I looked around for a GUI framework in Python and I was thrilled when I found : https://python-gtk-3-tutorial.readthedocs.org/en/latest/

With Glade + Python 3 I was chucking out GUI code faster than the Visual Studio + Win Forms because there was no compile step. No over engineered MVVM ala C#/WPF for a simple app - load a glade layout file, hook up callbacks - bam everything works together. The MVC is already there because the content pipeline (model) is already separated and views are defined in glade files. I had a prototype functional by the end of the day - it would take me that much just to choose which tools to use if I went Javascript client/Python server.

Seriously - couldn't believe how fast you could get stuff done with it - everything works as expected, you get all the controls you need - no need to learn random stuff like that qtscript, no build process (this is actually huge !), insanely fast iteration (faster than web-dev). I always thought of GTK like some obscure C gui lib but the Python wrapper makes it very usable.

My only complaint so far is that Glade seems to be getting progressively slower as you remove/add stuff so you need to restart it from time to time (eg. once per hour) but it starts instantly so it's not really an issue.

Note : I haven't yet tried to get it running on Windows or OSX since we all use Linux internally.

Pretty much this.

Anyone that has ever used RAD tooling for GUI applications can easily see how the HTML/CSS/JavaScript combo tooling is still miles behind from a 90's desktop developer experience.

Seeing you saying good things about Cordova makes me think you have never used Cordova.

I have, and I agree with what he said. One thing which isn't that great, are the plugins. I feel like there could be easier ways to access / write plugins. But overall, everything said was pretty spot on.

> Something based on Cordova (eg PhoneGap, Ionic) with Crosswalk can get you to about 90% of the performance of a native app

People always go on about Cordova etc, but I've never used one of these that wasn't a clunky, slow non-native looking mess. Do you have any examples of Cordova apps which aren't noticeably more crap than, say, an average-quality iOS app?

The best example I know of is http://snowbuddyhq.com/ ... It's pretty much impossible to tell that it's not a native app, it's fantastically well designed, and it works really well. There's a good article about it on Digital Arts: http://www.digitalartsonline.co.uk/news/interactive-design/w...

Installing this on Android, it's pretty obvious this isn't native:

* the very first animation (a slide transition to the left) was janky

* The app was unable to get my GPS location (it's being shared properly, and I have intermittent GPS signal, but if it was using the native fused location provider this would work perfectly)

* Buttons have no touch feedback.

* It's not the platform standard navigation drawer

* It's not the platform standard action bar

* The touch drag on the navigation drawer lags behind the finger position more than it should

* Non-native map has really bad pinch-zoom-pan gestures

As an Android developer I'm sure I'm consciously noticing things that lots of other users might not, but if this is the best Cordova still has to offer... :/

> As an Android developer I'm sure I'm consciously noticing things that lots of other users might not

Having talked to a lot of users, they do notice. They may not be able to articulate it, but all the things you mentioned give them the feeling that the app is not as good as some other app.

Great hybrid app but I can definitely tell it is not native. The animations stutter quite a bit on my nexus 6. I'd imagine it would be much worse on low to mid level android devices. Then there is that slight delay on all tap actions. Perhaps your average user wouldn't pick up on these things, but as a developer I definitely can spot a html app immediately and find them unpleasant to use in most cases.

Is a "win" for the developer necessarily a win for the users? Whose experience takes precedence?

If cost of developing app is halved, many more apps will be built. The choice is often not between "expensive native app" and "cheap hybrid app", it's between "why is it so expensive? but I can afford this (hybrid app)" and "no way that's too expensive (native app)".

In theory, this might be true. In practice, I find that it is just an excuse to cut corners and ignore the problems.

I recently had to help somebody with their public library and a particularly magazine publisher for digital issues. They require a web app to read the issues. They were trying to use their Mac, but the site refused to load. The public library staff had no clue why it didn't work ("well, it works for us", and their attempts to contact the publisher was your stereotypical tech support horror story.) Long story short, their website doesn't work on all browsers. They implicitly know that it doesn't work in Safari because they apparently wrote a native app just for iOS. However, they have no native app for Mac and never bothered to fix their website.

But is it better to have more apps, if most of those apps work like shit? Wouldn't it be better to have fewer apps, (although I reject the premise that it would happen) but higher quality apps?

If you need to get to market, then it could well be a win for both. If you then have significant traction and resources and still think 90% is good enough, well then, that's your risk to take as a business. Personally I feel that 90% is only good enough to get you started and there is a tipping point where user experience necessitates quality over business optimisations (shortcuts).

I don't think the tools we use impact the user experience that much, certainly not as much as the designs we use or the UX choices we make. If a hybrid app has a bad user experience it'd probably still be a bad experience if it'd have been built as a native app instead.

The article calls out scrollbar hijacking. That is a very example of the tools screwing up the user experience.

You generally don't have those problems in native. Apple in particular designs their APIs to steer you in the right direction and make it painful for you to do something that goes against their UX guidelines.

Yes and no.

One thing that is more and more important is batterylife. The more layers involved, the more energy they suck.

That's kind of old world thinking, it's been bugging me. In traditional programming and processor architectures that statement would be true. But if it were true today then Google wouldn't have built the Dalvik and then ART layer into the OS. They'd be compiling them on the developers PC and you'd end up with a fully static App.

The VM approach allows the VM to synchronise application wake-ups which can increase battery life.

Google chrome on Android is acting much like a VM to developers and I'm willing to bet the Android Chrome team are spending a much time as the Java runtime team are on optimising.

> Something based on Cordova (eg PhoneGap, Ionic) with Crosswalk can get you to about 90% of the performance of a native app

Companies like Famo.us have proven this is incorrect mathematically over 2 years ago. That is precisely why Famo.us built its own rendering engine that subverts the DOM and renderings content similar to a gaming engine (Unreal Engine). The DOM can't render as many surfaces as a native app in one view smoothly.

You can still make a very performant cross platform DOM based application (http://hn.premii.com is an excellent example), but in order to get that performance you need to limit your interface to very basic UI elements.

And even that site lags in animations. Something as simple as pull to refresh drops frames.

I'm using React Native as a web developer who wants to create a native app. Native performance with the familiarity and simplicity of web devlopment. I don't know why anyone would build a Cordova app anymore.

does react native support android yet?

About 4 months away now I think. But you could conceivably use something like Reapp until React Native supports Android. That's what I'll likely end up doing. Not ideal but React Native has already saved me so much time I'm not worried about the Android problem really.

Thanks for mentioning Reapp, I hadn't heard about it. I'm facing the same problem as you as I'm starting to build an app for Android and iOS and I can't really see myself building it the 'old way' after having tried React Native. Have you tried Reapp?

In my case it's very simple - so I can deploy the same code to the web. Not many people will want to do that though.

> building a hybrid app wins every time.

Define your criteria for "winning". If I'm truly building a user experience, then I don't know a single hybrid app that has "won" (my definition of winning here is an app I use daily, and I don't know of a single app I use daily that is hybrid)

> 90% is easily good enough for most applications

The problem is most applications (80%) don't really even need a native app to begin with.

crosswalk adds 58mb of weight on your app though and works on android only.

That's not been a problem for me, but if it is you can separate the x86 and Arm builds so it only half as much for each one.

I really dislike steam's hybrid approach.

"Something based on Cordova (eg PhoneGap, Ionic) with Crosswalk can get you to about 90% of the performance of a native app, with about 1/2 of the development time because it's all built with web technology."

This is true only if you're working with people without experience. Someone who is experienced with the mobile toolkits is going to be able to develop the app just as fast, with more functionality.

Reply on news.ycombinator.com