If you've ever tried to use the MySQL command line tool in some automated bash scripts or cron jobs, you'll probably wonder how to pass your password to MySQL without having to type it in. The most obvious choice would be to use the -p or --password command line arguments. If you've ever tried that, you'll probably have seen this warning message:

mysql: [Warning] Using a password on the command line interface can be insecure.

This is because any other users or programs running on that computer will have the opportunity to see the password, because process command line arguments are publicly available on Linux for some reason.

Today I learned a better way to do this. Actually I learned two ways from this Stack Overflow question:

1. You can use a configuration file, either in ~/.my.cnf or in a file specified with the --defaults-file command line argument, or
2. You can use the MYSQL_PWD environment variable.

For me, #2 was the easiest solution, even though it's not the most secure option to use generally. MySQL documentation warns:

This method of specifying your MySQL password must be considered extremely insecure and should not be used. Some versions of ps include an option to display the environment of running processes. On some systems, if you set MYSQL_PWD, your password is exposed to any other user who runs ps. Even on systems without such a version of ps, it is unwise to assume that there are no other methods by which users can examine process environments.

If this sounds scary, go with the configuration file approach. I'm not worried about environment variables on my servers, and this other Stack Overflow thread, Is passing sensitive data through the process environment secure?, explains that this used to be a big concern, but nowadays the MySQL documentation warning is a bit over-the-top. These days, on most Linux systems, environment variables are a safe way to pass around secrets.

Since I was already using environment variables to store my database credentials, I decided to just change my bash script from this:

#!/bin/bash

mysql -h "$MYSQL_HOSTNAME" -u "$MYSQL_USERNAME" "$MYSQL_DATABASE" --password="$MYSQL_PASSWORD" "$@"

to this:

#!/bin/bash

MYSQL_PWD="$MYSQL_PASSWORD" mysql -h "$MYSQL_HOSTNAME" -u "$MYSQL_USERNAME" "$MYSQL_DATABASE" "$@"

Sure, I could've renamed my environment variable from MYSQL_PASSWORD to MYSQL_PWD, but I have other code relying on that, so this was the easiest way to eliminate that warning.

Update November 2022: Why I love Mastodon. Follow me at https://toot.cafe/@JesseSkinner

I used to love Twitter. When I signed up in 2008, it was a new place to publicly talk with people from around the world. Over time, more people joined, and it became a great place to hear the latest news and gossip in the web development community. I enjoyed seeing famous people chatting and even roasting each other. It was exciting to hear rumours and see things unfold in real-time that would end up on the mainstream news. I'd spend a lot of time crafting tweets. Tweeting became part of my thought process, so that when something interesting in my life happened, I'd instantly start thinking of how to explain it in 140 characters.

Last year, that novelty wore off. I don't know if it was because of the increasingly divisive political environment in the USA, or the pandemic making everyone more stressed out and less patient, or if it was something that would've happened regardless. Every time I went on there, I'd come away with a bad taste in my mouth, angry or frustrated about one thing or another. I don't mean angry about injustices or world events, where the feeling is justified. I'm talking about being furious that somebody was wrong on the Internet!

Twitter started to remind me of one big comments section for the Internet, with everyone fighting about everything. People who I admired regularly get swept up in this hatred. One person vents their frustration over something, another tries to make a tasteless joke, others jump in to attack the replies, others jump in to defend, and on it goes.

I already stopped using Facebook years ago, and never got into Instagram or other platforms much, so Twitter had become the place I'd go on my phone when I had some spare time or just felt like zoning out for a while. I started to notice a pattern, that I would frequently tell my wife about things unfolding on Twitter, usually some big controversy that had everyone riled up. Whether I agreed or disagreed, I would get swept up in it. She was very patient to let me tell these stories, but would remind me that I don't have to share everything that happened on there.

The problem was, Twitter was taking up more and more of my thoughts and my mental energy. I would find myself thinking about these controversies and tweets in my free time, wondering if I should jump in with my opinion, but more often, I was just emotionally watching these things unfold from the sidelines, scared to be caught in the crossfire.

I finally decided to quit on Christmas morning. After opening presents from Santa and having Christmas breakfast, I habitually and unfortunately pulled out my phone and turned to Twitter for some heart-warming tweets. First thing I saw was a politician wishing everyone a happy holiday, and another politician replying to that with "Bite me". I couldn't believe it. I wanted to write back and call this politician out for being crass, tasteless and mean. But I realised there that then I would also be swept into this hate-fest as well, on a day where I should be focused on my family and having a lovely day. I decided right then that this would be the last time I read my feed on this hateful, toxic platform.

I decided that I don't need Twitter at all. I can't think of any real benefits I'd ever gained from using Twitter. I already have my blog, my newsletter and my YouTube and Twitch channels, where I can share my ideas and express myself. There's no good reason to continue to pour my energy into crafting content that ends up contributing to and supporting a private platform I don't even like.

The past few months have been wonderful. I installed a nice, simple blog reader called "Flym" on my phone, and subscribed to a bunch of web development blogs, some I had read on Google Reader back in the day, others were new to me. It's so much nicer to read longer, well-thought-out blog posts, that leave me feeling inspired, educated, and curious about the topics I'm interested in.

The one thing I'm kind of missing is the ability to write very short content to share a small thing I'm excited about or find funny. I figure that I'm better off to put that creative energy into blogging and writing newsletters more often, even if that means writing shorter posts, though probably more than 140 or 280 characters at a time.

Anyway, I just wanted to get off my chest all the reasons you hopefully won't find me active on Twitter anymore, and let you know that you'll likely find me writing blog posts on here more often.

Here are some other web development bloggers I've been enjoying lately, and I'm eager to find some more. If you also have a blog, I'd love to hear from you so I can subscribe to your RSS feed in my blog reader!

• Axel Rauschmayer
• Bruce Lawson
• Jake Archibald
• Jeremy Keith
• Julia Evans
• Marc
• Remy Sharp
• Sunil Pai
• Terence Eden
• Zach Leatherman

I've wanted to launch one of my side projects for a very long time. I'm the kind of guy who loves starting things, but never finishes them. Well, this week I finally finished one of them, by launching The Joy of Svelte, my first online video course!

Finding inspiration in an old five-year plan

Back in December, 2019 (a year ago, but feels like a decade), I got a new notebook for Christmas, because I had filled up my old one that I use for meeting notes, To Do lists, and stuff like that. I started re-reading my old one, and saw that near the start, I had a five-year plan from December, 2015. I had a goal to expand beyond my freelancing business and launch my first video course in 2016, with the goal of continuing to create courses, apps and other products over the coming years. By the year 2020, I wanted to have a whole catalog of courses and products under my belt.

Well, when I reading that, four years had passed, and I still hadn't launched anything. Not a single app or course, other than the dozens of free videos I'd recorded for YouTube, and, of course, all the client work I'd done as a freelance web developer. But I still wanted to do something for myself, something of my own creation.

The disappointment and shame I felt while reading this was the push I needed to finally commit to this lifelong plan and stick with one of my side projects long enough to actually see it through to launch. So, in January, 2020, I decided that I would focus on the things I was most excited about: teaching, recording videos, and Svelte.

Trying to stay motivated by committing publically

I had it all figured out. I put up a landing page that said "Coming in the Spring of 2020", and publically announced that I was working on a new course. I committed myself to my newsletter subscribers and Twitter followers, I put out an announcement on YouTube, and then hoped that all the public accountability would force me to follow through and finally launch something.

Well, that was all back in February. Spring came and went, and I was still stuck planning and trying to decide on the course contents. In June, I had to update the landing back and change it to "Coming Summer 2020..."

Coding as a form of procrastination

I decided early on to self-publish The Joy of Svelte by developing my own video course platform. I'm a web developer, after all, and it's way too easy to feel like I'm being productive when I'm writing code. So in a way, it was a form of procrastination, because I could sit down to integrate Stripe, or create a video player interface, or write code to deal with emailing out access links, and feel like I was making progress. In reality, I could've just used one of the many video course platforms available and saved myself a lot of time and effort.

Having now built all that out, I'm happy I did, because now I can self-publish more courses in the future. But I realise now that I could have launched a lot sooner if I had focused on recording videos and spent less time writing code.

Pivoting to focus on learning objectives

I started to record some videos, with the idea of making an SVG drawing app using Svelte. I recorded three videos showing how to do this, until I got to a point where it was starting to be more about SVG particulars and less about Svelte.

Eventually, I came across some very useful advice about creating course content: identify what it is specifically you want people to learn, then go and teach those things. I know that seems super obvious, but somehow I'd lost track of that and was instead accidently trying to make a course teaching people to make an SVG drawing app, but I don't think many web developers have the need to make SVG drawing web apps.

I looked at the landing page I'd originally made, and saw that I'd already outlined some key topics that I was planning to include:

You will learn about:

• getting started with Svelte
• using templating syntax to render data
• data fetching strategies for components
• using components to simplify complex web user interfaces
• building custom stores for state management
• integrating Svelte into an existing web application
• ...and more!

I decided to make six new videos, each one focused on one of these learning objectives. It was a simple, straightforward approach that ended up working very well, because it kept me focused on what it is I wanted people to learn, and less on what cool thing I wanted to build as a code example.

Back to the drawing board

So I abandoned the SVG drawing app videos, and started from scratch. I looked for some simple free web APIs and found one for Nobel Prizes, and decided that I'd use that to show people how to fetch data from an API. It needed very little explanation, didn't introduce any new, unrelated concepts, and more closely resembled the kind of API that I'd often used to build web interfaces for my clients. It might not be super fancy or flashy, but it allowed me to focus on Svelte instead, which is what mattered.

Off screen, I sat down and built a UI for browsing, search and filtering Nobel Prizes, to see if that would work well for the videos, and it turned out to be perfect. It gave me lots of different opportunities to demonstrate various Svelte features, and plenty of ways to show off what makes Svelte a joy to work with, and all the different strategies for making clean, reusable web components using Svelte. None of it felt contrived, all of it was applicable to real world web applications. I was ready to start recording.

Early access pre-launch and a final push to finish

Summer 2020 was coming to an end, and I did not want to change the release date on the landing page again. So, in one day, I sat down and recorded three of the six videos. I uploaded them to YouTube as unlisted videos, and on the very last day of summer, I sent out an email to my newsletter subscribers announcing that Early Access was now available.

It felt so good when I had my first sale ever! And then another one came! And then, while I was sleeping, another one! People were actually willing to pay me for my videos! This was a huge milestone in my life and career, and really validated all the work I'd put into it.

Still, I had three more videos to record to finish it up.

Benefiting from my own misfortune

Then something horrible happened. I recorded two more videos in one day, but when I finished, it turned out that OBS had used the wrong microphone, and so the audio was total garbage. I had to painfully decide to throw those videos out and re-record them.

Actually, that turned out to be beneficial, because I wasn't totally happy with some of my examples, and ended up coming with better examples that demonstrated the strengths and weaknesses of the different types of Svelte stores before I re-recorded the videos.

Launch day, and being too early

Three months after my Early Access launch, I had finally finished all the videos, and was ready to put the finishing touches on my web site, so that people could get a private link to watch videos directly on joyofsvelte.com instead of on YouTube. Using unlisted YouTube videos had felt a bit unprofessional, although I don't think anybody would have complained if I had stuck with using them.

Finally, on Monday, December 14th, 2020, I launched my first video course ever! I created a promotional video, and posted it with an emoji-filled tweet to Twitter.

On launch day I had two sales, and woke up the next morning to a third sale. I had tempered my expectations so that I wouldn't be disappointed, and so I was actually pleasantly surprised to make any sales that day. I'd figured most people who were excited about the course would have bought it during Early Access, and that turned out to be mostly true.

I also have come to realise that I'm probably way too early to be launching a course about Svelte. I chose Svelte because I'm so excited about it, and am happy to talk about Svelte endlessly, but the fact is, Svelte is not yet widely adopted amongst web developers, so there really isn't a huge audience there yet. It's still somewhat a niche topic. And that's okay, but it means that there was no way I was going to have a ton of sales on the first day.

There just aren't that many people learning Svelte right now. I think this will change over the coming months and years, and I'm glad to have put this course out into the wild to help people looking to learn Svelte. I hope it helps people to see what it is about Svelte that I find exciting, and why it has changed the way I approach web development altogether.

Lessons learned for the next course

This won't be my last course, it's just the beginning. Here are some lessons I've learned from building this course, that will change the way I approach building my next video courses.

1. I'll focus on learning objectives from the start. I'll make a short list of what I think people will want to learn about, and make videos focused on those points. The code examples I use will be chosen for how well they can demonstrate those key learning objectives.

2. I'll avoid perfectionism, and limit how much time I spend planning the course up front. Planning is a trap that I fell into, because you can keep planning the same thing forever. At some point you have to say "good enough" and start doing the actual work. Chances are, when you actually start recording the videos, you'll find out the best way to do things.

3. I ended up re-recording a lot of The Joy of Svelte by accident, and that benefitted me by allowing me to improve the content before recording the final videos. I will do this on purpose next time, maybe live streaming the content on Twitch, or possibly running a workshop beforehand, so hopefully I can get some useful feedback first as well. (And I'll try to remember to double check my microphone before recording the final videos!)

On to the next side project

One of the best things about launching The Joy of Svelte, is that I can now start working on all the other side projects and ideas I came up with this year, but wouldn't allow myself to work on until the course launched. If you're interested in following along, you can sign up for my newsletter.

And, of course, if you're interested in learning Svelte, check out The Joy of Svelte!

In case you missed it, Rich Harris gave a presentation at Svelte Summit 2020, where he announced that Sapper v1 will never be released! Instead, he showed what's coming next in Svelte itself.

Be aware that at the time of me writing this blog post, none of this is officially released yet, and very likely will change in the near future. Nonetheless, it's exciting to see a sneak preview of what the future of Svelte will look like.

Getting started

To get started today, you can run this command in your terminal, assuming you have npm installed: npm init svelte@next

In the future, it'll probably just be npm init svelte, which is super clean and easy to remember. This will be a nice change from having to run npx degit svelte/template my-template.

Here's what you'll see if you run this command today:

█████████  ███████████    ███████    ███████████  ███
███░░░░░███░█░░░███░░░█  ███░░░░░███ ░░███░░░░░███░███
░███    ░░░ ░   ░███  ░  ███     ░░███ ░███    ░███░███
░░█████████     ░███    ░███      ░███ ░██████████ ░███
░░░░░░░░███    ░███    ░███      ░███ ░███░░░░░░  ░███
███    ░███    ░███    ░░███     ███  ░███        ░░░
░░█████████     █████    ░░░███████░   █████        ███
░░░░░░░░░     ░░░░░       ░░░░░░░    ░░░░░        ░░░

Pump the brakes! A little disclaimer...

svelte@next is not ready for use yet. It definitely can't
run your apps, and it might not run at all.

We haven't yet started accepting community contributions,
and we don't need people to start raising issues yet.

Given these warnings, please feel free to experiment, but
you're on your own for now. We'll have something to show
soon.

It'll go on to ask you if you want to use TypeScript, which is really nice for those who like to use TypeScript, and nice that it's optional for those who don't.

Here's the full directory structure you will get with an initial installation:

├── .gitignore
├── package.json
├── README.md
├── snowpack.config.js
├── src
│   ├── app.html
│   ├── components
│   │   └── Counter.svelte
│   └── routes
│       └── index.svelte
├── static
│   ├── favicon.ico
│   └── robots.txt
└── svelte.config.js

Starting the dev server

Once it's done setting up files, you need to run npm install and then npm run dev to spin up the dev server. Here's what you'll see:

snowpack

  http://localhost:3001 • http://10.0.0.180:3001
  Server started in 643ms.

▼ Console

[snowpack] installing dependencies...
[snowpack] ✔ install complete! [0.59s]
[snowpack] 
  ⦿ web_modules/                                size       gzip       brotli   
    ├─ svelte-hmr/runtime/hot-api-esm.js        22.08 KB   7.4 KB     6.29 KB    
    ├─ svelte-hmr/runtime/proxy-adapter-dom.js  5.17 KB    1.65 KB    1.38 KB    
    ├─ svelte.js                                0.18 KB    0.15 KB    0.11 KB    
    ├─ svelte/internal.js                       52.36 KB   13.16 KB   11.36 KB   
    └─ svelte/store.js                          3.3 KB     1 KB       0.88 KB    


[snowpack] > Listening on http://localhost:3000

What is happening under the hood? This is very different from the Svelte and Sapper templates that came before. There is no longer a rollup.config.js nor a webpack.config.js, because it does not use Rollup nor Webpack, at least not during development.

Instead, it uses Snowpack to handle compiling and serving client-side resources. Snowpack does not bundle your resources, and relies heavily on JavaScript’s native module system, which means development is much faster. There is even a snowpack.config.js file which gives you a place to configure Snowpack to some degree:

// Consult https://www.snowpack.dev to learn about these options
module.exports = {
    extends: '@sveltejs/snowpack-config'
};

Building your application

There is now also a new svelte.config.js file, which lets you define an "adapter", used with npm run build to build your application into a production web site:

module.exports = {
    // By default, `npm run build` will create a standard Node app.
    // You can create optimized builds for different platforms by
    // specifying a different adapter
    adapter: '@sveltejs/adapter-node'
};

The default adapter will use Rollup to build your site into a Node.js web server. It seems that this web server doesn't use Express.js, though that might change as well, or maybe there will be a special adapter for Express.

If you want to have a purely static export, you can currently replace @sveltejs/adapter-node with @sveltejs/adapter-static, but be sure to run npm install @sveltejs/adapter-static as well.

In the future, there will be many other adapters, for example, building specifically for certain web hosting platforms, serverless architectures, and who knows what else? The cool thing about this adapter approach, is that you can build your web site without necessarily knowing how it will be built or deployed. You'll be able to change the adapter without changing your code.

Dependencies

Let's have a look at the package.json:

{
    "name": "demo",
    "version": "0.0.1",
    "scripts": {
        "dev": "svelte dev",
        "build": "svelte build"
    },
    "devDependencies": {
        "@sveltejs/adapter-node": "0.0.12",
        "@sveltejs/kit": "0.0.23",
        "@sveltejs/snowpack-config": "0.0.4",
        "svelte": "^3.29.0"
    }
}

Note that there are very few dependencies here. I really like how minimal this is. Both of the scripts are using the new svelte CLI from @sveltejs/kit, though that name might change, and it's not even available on GitHub yet. For now, you can look at the npm package.

Routes

You'll notice a folder src/routes/ where you can define your routes similar to how Sapper (or Next.js, etc.) let you define routes. Basically, your folder and file structure in here will map one-to-one with the routes on your web site. This is really nice, and easy to work with, especially if you're used to using PHP or other similar web development platforms.

If you're not building a static-only web site, you can also define server-side routes, similar to what you can do with Sapper. For example, you can create a file at src/routes/api.js:

export async function get(req) {
    return {
        status: 200,
        body: {
            hello: 'world'
        }
    }
}

If you're familiar with Sapper, you might notice that you have to return an object with status and body properties, instead of using an Express res object for your response. This is because it is not Express middleware. It uses an internal Node web server, with an API similar to what you may have used with some serverless cloud functions.

To create a layout component, to provide a consistent header and footer wrapped around all your routes, you can create a file called $layout.svelte, similar to Sapper's _layout.svelte.

You can also make an error handler route called $error.svelte, to handle 404s and other programming errors. It receives a status prop and also an error prop, so you can decide how to display the error to your users.

Migrating

Rich Harris notes that migrating from Sapper or other similar frameworks should be fairly straightforward, since most of the folder structure and other concepts are pretty similar. You'll probably just need to rename some files, and change how your server-side routes work, because they will no longer be written as Express middleware.

For fetching data for both server-side and client-side rendering, the Sapper approach of having a <script context="module"> block currently still works, though it's possible that will change.

Conclusion

If you're excited about all this stuff, it's definitely too early to start building your applications using it, but I'm willing to bet that it'll be a good choice to get started by using Sapper today, with the expectation that it'll be easy enough to migrate to this in the future, once it's ready.

To see a demo, check out Rich Harris' video Futuristic Web Development

If you're interested in learning more about Svelte, check out my video course The Joy of Svelte.

In many ways, time has been going very slowly amidst this pandemic, as we've all waited for case numbers to come down, lockdowns to lift, and vaccines to be tested. In other ways, time has been flying by. It's hard to believe I bought joyofsvelte.com almost seven months ago, and my initial Spring 2020 launch date is already months behind us.

I've never launched a product on my own before. I didn't realise how different it would be recording videos for a paid video course compared to making free videos for YouTube. If I'm going to expect people to pay, these videos have to be absolutely perfect. That means when I record them, I need to have a large block of uninterrupted time, completely free of distractions, and I need to be well-rested, in a good mood, full of joyful energy and inspiration.

Well, I have a four-year-old at home. I absolutely love being with my son all day, but there's definitely not much stillness around here, and any extra energy is either spent on getting some freelance work done, cleaning up the house, or trying to fit in a grownup movie or game before bed.

Last weekend, my wife took my son to a socially distanced birthday party, which marked the first time I had the house to myself in four months. I got to sleep in too, so I was well-rested, in a quiet house, full of joyful energy and inspiration. It was time to finally record. And I did! I recorded two more videos. Well, one and a half before they came back home. But I had momentum, so while we let my son quietly play Mario Kart, I went and finished the second video!

It's hard to find time for side projects. It's even harder when you have such high standards for that time. They say that perfect is the enemy of good, and I'm coming to realise that something like a video course can never be perfect anyway. I can only do my best with the time and energy I have available. I'd rather finish this course and share my experience and insights on using Svelte with the world, than to plan it forever and never launch.

I wrote that tweet back in May and obviously I've failed to take my own advice. But I'm not giving up. I'm going to keep at it and keep trying to find the time. We're probably going to keep my son home from school this fall, so it doesn't look like I'll have a quiet house for a long while. I'm going to have to get creative and find a way to make it work.

I hope you're able to find some time for your own side projects too, and know that you're not alone in that struggle. We usually only see people launching projects once they're already done. I'm sure there are countless more unfinished and unlaunched side projects that the world will never know about. Don't let your side project become one of them.