DONN FELKER

Lessons Learned From the Software Industry

Connect

Why Mobile is Not the Future …

by

Why Mobile is Not The Future - Donn Felker

That’s right, I don’t feel that mobile apps are the future. I’ve been saying this for years.

As someone who has spent the last decade plus working solely in the mobile space on Android, this is a hard realization to accept. Hell, I’ve even written books about mobile development and how to make apps. Some of my businesses surround this topic and the majority of my consulting business is mobile … so why am I saying this? Have I lost it?

Before you make any snap judgements about my prediction, please hear me out…

Let me ask you this…

What app CAN YOU NOT live without on your phone?​

You might say “Instagram”, “TikTok” or “Twitter”, “Email”, or something similar …

Wrong.

The web browser. ​

Why?

It’s the ultimate integration platform.

When you don’t have an app installed, what happens? You’re taken to a url in a browser and you interact with the website in which the url points to.

What if you don’t have a browser installed? Your phone, as a data device, is nearly useless. Sure, you can use some apps. But now let’s take the web views out of those apps. My guess is that over 60-80% of them will fail – if not more.

I bet that one of the most popular apps on your phone is your Email App.

For the sake of simplicity I’m going to assume you use Gmail.

Did you know that Gmail App is mainly a WebView?

It’s true.

Open any email in the Gmail App and zoom in (two finger pinch to expand) and you’ll notice it zooms in (try it on this one if you’re reading this in the Gmail mobile app). The zoom you see is the WebView in action. I’ve spoken to the folks who work on this app at Google and they were the ones that told me it was mainly a WebView. I was shocked at first. 🤯. Literally blew my mind a few years back.

One of the most popular apps in the world … is a WebView. Sure, there are some native components, but … the main component … is a WebView.

Think about it. Almost every app I work with has some web view in it somewhere. Take the web views away (and browsers) and BOOM, things don’t work.

However, don’t be dismayed.​

Android and iOS will have highly valuable purposes for the foreseeable future, though.

Many apps are far better as native apps. Such as media apps like video and audio editors. Social media apps like Facebook, Instagram, Twitter, TikTok and more. Maps and Games will need to be native as well and some other apps will too. It really depends on the use case.

​

The Next 10 Years …

I predict that in the next 10 years we will see a surge of web applications and web technology that enable us to run near native or faster than current native technology in our mobile and desktop browsers.

Many are going to say “everyone has been saying this for years and it hasn’t happened, it’s not going to happen.”

How do you know that? In my opinion, the writing is on the wall.

How will the web take over?

Browsers will get faster and better at what they do. They’ll have more features, etc. Data speeds are going to increase and it will be easier to download more data/web apps quickly.

Things like web assembly will have frameworks written for them in high level languages that we can easily code to (or new high level languages will be created). The devices will expose more of their native technology to the browser so developers can easily access it (more bluetooth, location, sensors, etc).

Eventually, storage, photos and system services and interactions will seem native on a web browser more-so than they already are.

It’s just how it will be.

When that happens, many apps will only need to be written once. For the browser, on the web.

HTML, JavaScript and CSS are cross-platform by default. They target the web browser that runs everywhere already.

Ultimately what will happen is … the browser will become the platform. ​

This is already happening and has been for years. ​

Look at Chromebooks. They run one thing …. CHROME. It’s a web browser in a laptop form.

If you have kids, you probably have a Chromebook or two at home right now for their schooling.

The writing is on the wall folks, I’m just reading it to you.​

Does that mean you should stop doing Android or iOS development today?

Not at all.

We’re not there yet. We’re not even in the infancy. We’re in the prenatal stage of this right now.

Things are still forming. We’re not sure what it’s going to look like, but we need to be aware of what’s going to be birthed in the future.

My gut and my bet is that the web will be the future.

Discuss this on Twitter

—

PS: I’m not telling you to abandon what you’re doing now if you’re a mobile developer. I just want you to be aware of the changes on the horizon that I’m seeing. If you’re interested in web development I advise you dive into a framework you find interesting. Such as Ruby on Rails, Python, Node.js (with Express), ASP.NET, Ktor (Kotlin based), or anything for that matter. Just tinker around with it. Build something with it. Having the skillset now will only help you in the future.

Did you know that you can write web applications with the Kotlin Programming language? I have a FREE 9+ hour course on the Kotlin Programming Language here. Its 100% FREE. Learn Kotlin and then dive into any number of Kotlin based web frameworks and you’ll be able to start with the web real easily.

Lastly, if you like content like this and want to receive it before anyone else does sign up for my email newsletter below. I send this content out to my newsletter a week to two weeks before it goes live on my site.

Code Comments

by

Code Comments

When it’s it appropriate to comment code?

I follow a simple set of rules:

  1. If the code is confusing – I try to refactor the code so that it’s not confusing anymore.
  2. If a refactoring is not possible, I document the “why”, not the “how”. The “how” is already documented: that’s the code.

The “why” explains WHY the code exists and perhaps some important and relevant details that might not be evident to future maintainers.

How do I know if I should provide a “why”?

I ask myself this question:

In six months will I be able to understand this code in 30 seconds?

If there is any doubt in my mind then I know that I need to provide a comment for that chunk of code.

Why 30 seconds?

The majority of my time as a developer will be spent reading and deciphering existing code. I need to be able to do this as quickly as possible to be efficient. If a comment helps me grok something quickly due to some weird edge-case complexity, then the comment is worth its weight in gold.

Dagger Open JDK Version Mismatch

by

While working on a client project I recent received this error:

An exception occurred: java.lang.IllegalArgumentException: not a valid name: <set-?>Provider

Googling for the error didn’t return any results, so I’m posting the fix here that I got from Philippe Breault (thanks Phil 👍).

There have been some issues filed that match this error here on the Google issue tracker.

The Problem

This error happens because of a version mismatch in Open JDK that is used in the terminal vs the one that is installed in Android Studio.

The Solution

To fix this, you’ll need to set the JDK Location to what the terminal is using.

To find what JDK path your terminal is using (on a Mac), run this command: /usr/libexec/java_home

For me it looked something like this (don’t copy this, its unique to your machine, this is just an example):

/Library/Java/JavaVirtualMachines/jdk1.8.0_221.jdk/Contents/Home

Once you have that you’ll want to open your Android Studio project which is giving you trouble and then press cmd+;to open the Project Structure window:

Android Studio SDK Location

You’ll want to update the JDK Location above to represent the value that was returned from the terminal ouput (the JDK path above) .

Once you do this I advice cleaning your project and invalidating caches on Android Studio and then rebuilding. At that point your build problems should be alleviated.

Enjoy.

How To Fix: Jest Did Not Exit

by

I was recently doing some work on a Node.JS project and I decided to use Knex.js with Objection.js for my data access to PostgreSQL.

Problem

In usual fashion, I wrote some tests to validate the behavior of the application and while writing the tests in Jest I found the tests were not exiting. I was getting the following error:

Jest did not exit one second after the test run has completed.

This usually means that there are asynchronous operations that weren’t stopped in your tests. Consider running Jest with --detectOpenHandles to troubleshoot this issue.

I ran the tests with --detectOpenHandles and the tests just hung. Nothing happened.

Solution

After some experimenting, I realized I was not destroying my Knex instance and that was keeping the tests from finishing.

In the afterAll() method of Jest I’m now destroying the Knex instance so that the test can finish appropriately.

Here’s a very naive implementation that works for me as an example:

The model – Person.js:

'use strict';

const { Model } = require('objection');
const bcrypt = require('bcryptjs');

class Person extends Model {
    // Table name is the only required property
    static get tableName() {
        return 'persons';
    }

    async $beforeInsert() {
        this.created_at = new Date().toISOString();
        await this.updatePassword();
    }

    async $beforeUpdate(opt, queryContext) {
        await super.$beforeUpdate(opt, queryContext);
        this.updated_at = new Date().toISOString();
        await this.updatePasswordIfChanged(opt.old);
    }

    async updatePassword() {
      // not prod quality, do not copy, example only
      // bcrypt stretches should be much longer
        const hash = await bcrypt.hash(this.password, 1); 
        this.password = hash;
    }    

    async updatePasswordIfChanged(old) {
        if (old.password) {
            const passwordMatches = await bcrypt.compare(this.password, old.password);
            if (!passwordMatches) {
                await this.updatePassword();
            } 
        }    
    }

    static get jsonSchema() {
        return {
            type: 'object',
            required: ['firstName', 'password'],
            properties: {
                firstName: { type: 'string', minLength: 2, maxLength: 255 }
            }
        }
    };
}

module.exports = {
    Person
};

The Test:

const Knex = require('knex');
const knexConfig = require('../knexfile');
const { Model } = require('objection');
const bcrypt = require('bcryptjs');
const { Person } = require('../models/Person');

const knex = Knex(knexConfig.test);

Model.knex(knex);

afterAll((done) => {
  knex.destroy();
  done();
});


test('can update password when a user is already saved', async () => {
  const person = await Person.query().insert({ firstName: 'John', password: 'foo' });
  const passwordMatches = await bcrypt.compare('foo', person.password)
  expect(person.password).not.toBe('foo');
  expect(passwordMatches).toBeTruthy();

  const updatedPerson = await person.$query().patchAndFetch({ password: 'bar' });
  expect(person.password).not.toBe('bar');
  expect(person.password).not.toBe('foo');

  const updatedPasswordMatches = await bcrypt.compare('bar', person.password);
  expect(updatedPasswordMatches).toBeTruthy();
});

Now the test will pass and it will not get hung.

PS: Yes, this is a simple example and the code is not what I’d put into production, so please ignore the content there – this is just an example of how to fix the “Jest did not exit one second after the test run has completed.” issue.

Fix: Deploying Docker Instances to AWS Elastic Beanstalk without Default VPC

by

Docker Kubernetes

File this under “I hope this helps someone else in the future“.

I was learning Docker and Kubernetes recently and I was trying to get my docker instance running on Amazon AWS Elastic Beanstalk on the us-east-1 region.

I had to configure my beanstalk environment to use a t2.small instance because the npm install command on the t1.micro was timing out.

Upon upgrading and pushing a new version to Travis CI (flow was Git -> Travis CI -> S3 -> Beanstalk) I’d get these errors:

  • Environment named Docker-env is in an invalid state for this operation.
  • Auto Scaling activity failed 5 minutes ago with error: The specified instance type can only be used in a VPC.
  • Message:The security token included in the request is expired
  • Service:Amazon S3, Message:The provided token has expired.
  • A search on the web wasn’t that useful for any of these errors, so hopefully this helps someone by my including them here. Here’s how I fixed this …

The Problem

I didn’t have a default VPC for the us-east-1 region. I needed to create one. (If you accidentally deleted the AWS Amazon Default VPC – see below for how to set it up again).

My problem was that my account is a very old AWS account (I’ve had it forever – set it up probably in 2008-ish) and my us-east-1 did not have a default VPC (needed for a t2.small). I could not create a default VPC (under the ‘Actions’ menu in the AWS VPC section for us-east-1) because my account in us-east-1 was a EC2-Classic and it needed to be an EC2-VPC. Long story short, regions have certain configurations and my configuration was set to EC2-Classic and that was causing a mess of problems.

The Fix

If you can create a Default VPC, do so. There is a link below to show you how to do that. If you can’t … read on …

To fix this I had to submit a support request to Amazon to ask them to upgrade my us-east-1 region to EC2-VPC from an EC2-Classic configuration.

To do that you’ll create a support request (while logged into AWS).

  • Open a new support case in the support console
  • Regarding: Account and Billing Support
  • Service: Account
  • Category: Convert EC2 Classic to VPC

Then fill out your request (ask them to upgrade you to EC2-VPC for that region) and you should be good to go.

A Workaround

In order to keep continuing on my local developments while Amazon converted me to EC2-VPC, I changed regions to Ohio (us-east-2) and recreated a new beanstalk environment and application. This region almost always has a default VPC. If you deleted it on accident, see the link below for how to re-enable the default VPC on that region.

Once your new AWS Beanstalk environment is set up, go into the configuration and change it from t1.micro to t2.small.

Wait for the changes to take effect.
Redeploy.
Presto, docker instance is deployed and I can access the web app via the beanstalk url.

Links

Photo:
unsplash-logofrank mckenna