Have some thoughts? Send me an email.

The problem is that the kids play on a PlayStation 5 and a Nintendo Switch, while we parents play on Mac computers, and Minecraft doesn't support cross-play between the Bedrock (consoles) and Java Edition (Mac/PC).

Enter Geyser

Geyser is a plugin for the Java Edition that allows Bedrock Edition clients to join the Java Edition server. See their Setup Instructions for how to install it. I did a quick evaluation/comparison of the listed hosting providers and found that CloudNord hosting has built in support for Geyser.

I've been a very happy CloudNord customer for my Minecraft hosting. If you're looking for Minecraft server hosting, you can use my affiliate link to sign up - and I'll receive some hosting credits.

You will also probably need to install the Floodgate extension; it handles some account-specific things between Bedrock and Java Edition players.

Connecting to the Server

It is still a bit complicated to connect to a Java server from Switch/PlayStation when Geyser is up and running, but there's a nice solution for that as well: the MCXboxBroadcast Geyser extension. That uses an Xbox Live account, and every friend of that account will see it as online on the server, which makes it easy to join the server from both Switch and PlayStation. I created a new dummy account for this purpose and made sure to befriend everyone who should be able to join the server.

Have some thoughts? Send me an email.

I decided to do something similar for this site, so now each post includes webmention statistics instead of the already existing stats from the Fediverse.

I made one small change to the webmention fetch function by combining mention-of and in-reply-to numbers.

Have some thoughts? Send me an email.

Have you ever wondered why the electric door on the passenger side of the Volkswagen Sharan won't open after you've fitted a new battery? Neither had I until last week when I picked up my car from the mechanic and noticed that the door button was blinking, and nothing happened when I pressed it. The closing mechanism was working fine after I, with quite some force, opened the door manually.

I asked the shop for help, but they were clueless and tried to diagnose the issue by connecting their iPad to the car's diagnostic port and running a series of tests with no luck. While they did that, I decided to do a quick search online and found a potential solution on a forum. The solution is to reset the automatic window setup on the door that won't open.

The solution

1. Using the window button on the door that won't open, open the window fully.
2. Release the button and press it again for 3 seconds.
3. Close the window fully.
4. Release the button and press it again for 3 seconds.

I did this once, and the blinking stopped, and the door opened when I pressed the door button. It was a weird fix, but it worked!

PS. I do like long blog post titles.

Have some thoughts? Send me an email.

Have some thoughts? Send me an email.

You can enable Standard formatting with the built in ruby-lsp extension in Zed. Just add this to your config.json:

"languages": {
  "Ruby": {
    "language_servers": ["ruby-lsp", "!solargraph", "!rubocop"],
    "formatter": "language_server"
  },
},
"lsp": {
  "ruby-lsp": {
    "initialization_options": {
      "enabledFeatures": {
        "diagnostics": false,
        "formatting": true
      },
      "formatter": "standard"
    }
  }
}

Zed and Standard Ruby

I’ve been using Zed as my main editor for the last two weeks or so. It feels great, lean, and fast compared to VS Code that I’ve been using for years. Last time I tried Zed, there were no possibilities to install extensions or add/customize language servers, but this has changed, and there are more and more extensions being built.

I missed one particular feature while working with Ruby code, formatting with Standard ("…a linter & formatter built on RuboCop…") so I set out on a mission to try to create one, how hard can it be?

Turns out it was a bit harder than I thought but not too bad. You have to write Rust, which I had never done before. Luckily, there were a lot of extensions to look at, and with that, I could quite quickly throw together a working extension that wraps Standard’s language server and uses that for linting and formatting.

Introducing the Standard extension for Zed

Github: himynameisjonas/zed-standardrb

The implementation is quite naive, and I'm not sure if I should submit it to Zed's extension library. We'll see. But until then, you can clone/download the code and use the "install dev extension" feature to use it yourself.

To enable it, you can add this snippet to your config.json:

"languages": {
  "Ruby": {
    "language_servers": ["standardrb"]
  }
},

Feel free to contribute or give feedback. I'm sure there are a lot of improvements that can be made, but for now, it works for me, and I have a huge lack of time to work on it further. I hope it can be useful for someone else as well.

Have some thoughts? Send me an email.

The "logo"

The "logo" on this site that is set in a custom font. It felt very wasteful to load the entire font just for displaying my name. Using the following command (I used Docker to run the tool to avoid installing all dependencies on my machine), I shrunk the font file from 79KB to just 6.5KB.

docker run --rm -v "./:/app" talentplatforms/glyphhanger --whitelist="JONAS BRUSMAN" --family="Tropi Land" --subset=TropiLand.ttf --formats=woff2,woff,woff-zopfli --css

I ran the command in the directory where the font file is located. The --whitelist flag specifies the characters I want to include in the subset. The --family flag specifies the name of the font. The --subset flag specifies the name of the output file. The --formats flag specifies the formats I want to generate. The --css flag generates a CSS file with the @font-face rule that I copied and pasted into my site's CSS file.

The main font

The main font on this site is also a custom font. I used the following command to shrink the font file from 318KB to just 32.7KB.

docker run --rm -v "./:/app" talentplatforms/glyphhanger http://host.docker.internal:8080/texts/ http://host.docker.internal:8080/photos/ --jsdom --US_ASCII --family="Overpass" --subset=Overpass-VariableFont_wght.ttf --formats=woff2,woff,woff-zopfli --css

This time I used the --jsdom flag to make the tool visit the URLs I specified to find the characters used on the site. I ran a local server (using Eleventy) on port 8080 and pointed the tool to the URLs where the text is located. The rest of the flags are the same as in the previous command.

Have some thoughts? Send me an email.

Some more or less notable changes

• New colors which I found on @randomColorContrasts@botsin.space.
• Trying out Open Heart Protocol buttons on each post, a tiny heart button to show appreciation for the post. Found out about it from benji.
• Removed Inter and replaced it with two new fonts:
  • Added Tropi Land as the custom font for the "logo" to give it a more playful look.
  • Added Overpass as the main font for the site.
  • Used Glyphhanger to subset the fonts to only the characters used on the site. Overpass went from 310.95 KB to 31.98 KB and Tropi Land from 77.11 KB to 6.42 KB.
• Renamed Notes to Articles, not sure if that will stick but notes sounded too much like short posts. Maybe it will come back, but as a syndication of the notes I post on Mastodon.
• Sprinkled some icons from Heroicons to the main elements/navigation.
• Replaced Instantpage with Quicklink for prefetching links.
• Setup Vite for js bundling with eleventy-plugin-vite.

Have some thoughts? Send me an email.

I suppose it will include links related to my interests, such as web development, programming in general, photography, the IndieWeb, and other random things.

Have some thoughts? Send me an email.

I connected my GitHub repository to Cloudflare Pages and had Cloudflare build and deploy the site. It worked, but was really slow due to the large photos it has to process every time it rebuilds. It took over 14 minutes to build and deploy the site 😩. Cloudflare supports build caching, but only for a few specific frameworks, and unfortunately, Eleventy is not one of them.

I remember reading about how Sophie Koonin deploys her site to NeoCities using GitHub Actions (go read that; it is a really good write-up of how things work). Knowing that GitHub Actions has support for custom caching, I decided to do the same but with Cloudflare Pages instead of NeoCities.

Deploy to Cloudflare Pages with GitHub Actions

I created a GitHub Actions workflow that caches the node_modules and .cache directories, as well as the img directory where the photos are stored. This reduced the build time to just over 2 minutes, a 7x improvement!

The full workflow

This is the complete workflow as I write this post. I may have updated it since then, so please refer to the source code for the latest version.

name: Build and Deploy

on:
  push:
    branches: ["master"]
  pull_request:
    branches: ["master"]
  schedule:
    - cron: "17 8 * * 6"

jobs:
  build:
    name: "Build and Deploy"
    runs-on: ubuntu-latest

    steps:
      - uses: actions/checkout@v3
      - uses: actions/setup-node@v4
        with:
          node-version: 20.10.0
          cache: "yarn"
      - run: yarn install --frozen-lockfile

      - uses: actions/cache/restore@v4
        id: cache
        with:
          path: |
            .cache
            img
          key: 11ty-${{ runner.os }}-${{ github.run_id }}
          restore-keys: |
            11ty-${{ runner.os }}
            ${{ runner.os }}-11ty

      - name: Build 11ty site
        run: yarn build
        env:
          ENABLE_ANALYTICS: ${{ github.ref == 'refs/heads/master' }}

      - name: Publish to Cloudflare Pages
        uses: cloudflare/pages-action@v1
        with:
          apiToken: ${{ secrets.CLOUDFLARE_API_TOKEN }}
          accountId: 8dfec48c278d5a97c93de33e072261bd
          projectName: jonas-brusman-se
          directory: dist
          gitHubToken: ${{ secrets.GITHUB_TOKEN }}
          branch: ${{ github.head_ref || github.ref_name }}
          wranglerVersion: "3"

      - uses: actions/cache/save@v4
        with:
          path: |
            .cache
            img
          key: 11ty-${{ runner.os }}-${{ github.run_id }}

The workflow does the following:

1. Checks out the code and sets up Node.js with a specific version, enabling caching for dependencies installed with Yarn.
2. Installs dependencies using Yarn.
3. Restores cached directories for faster builds. I use the unique github.run_id to create a unique key for the cache so it can be updated with new content/images on every build.
4. Builds the 11ty site and sets some environment variables for the build.
5. Publishes the site to Cloudflare Pages using the Cloudflare Pages Action, specifying project details and authentication. I had to specify the branch name manually with branch: ${{ github.head_ref || github.ref_name }} to make it work with pull requests.
6. Caches the contents in the specified directories to speed up future builds.

Have some thoughts? Send me an email.

After deploying Jemalloc, our app's memory usage decreased from around 6 GB to just 2.5 GB. Super nice!

Our next step involved switching to ARM architecture with AWS Graviton for our ECS tasks, resulting in improved performance at a lower cost. Two big wins in a short time!

Have some thoughts? Send me an email.

Our Docker image setup for jemalloc, based on a gist from John Bachir, was using a hard-coded path containing a reference to x86_64:

ENV LD_PRELOAD=/usr/lib/x86_64-linux-gnu/libjemalloc.so.2
ENV MALLOC_CONF="dirty_decay_ms:1000,narenas:2,background_thread:true"

The solution I came up with to support both linux/amd64 and linux/arm64 was to use a symlink:

RUN ln -s /usr/lib/*-linux-gnu/libjemalloc.so.2 /usr/lib/libjemalloc.so.2
ENV LD_PRELOAD=/usr/lib/libjemalloc.so.2
ENV MALLOC_CONF="dirty_decay_ms:1000,narenas:2,background_thread:true"

Have some thoughts? Send me an email.

The big win

That's when I found Migadu. They have a different billing approach that works better for families like mine. Instead of charging for the number of email addresses or accounts, Migadu’s pricing depends on the volume of emails sent and received. This means I can create as many accounts, addresses, or aliases as we want without worrying about the cost.

The missing stuff

Migadu doesn't offer calendar or contact syncing, but I solved this by moving these functions to Apple iCloud, which is fine since I'm already in the Apple ecosystem.

Have some thoughts? Send me an email.

📨 Mail Service: Migadu

📮 Mail Client: Apple Mail

📝 Notes: Apple Notes & Logseq

✅ To-Do: Things

📆 Calendar: Apple Calendar

🗞️ RSS service: Feedbin

⌨️ Launcher: Raycast

🛹 Clipboard manager: Raycast

☁️ Cloud storage: iCloud

🌅 Photo library: iCloud

🤳🏻 Photo editing: Darkroom

🌐 Web Browser: Orion for browsing, Firefox for web development

💬 Chat: Signal

🎤 Podcasts: Overcast

🎶 Music: Apple Music

🔐 Passwords: 1Password

💁🏻‍♂️ Social: Mastodon

🐘 Mastodon: Ivory

🌤️ Weather: Carrot Weather

🔎 Search: Kagi

🎮 First game I play each morning: Threes

👨🏻‍💻 Terminal: iTerm2

🧮 Code Editor: Visual Studio Code

🛍️ Grocery List: Matlistan

Have some thoughts? Send me an email.

The setup is based on Datadog's article on monitoring Ruby on Rails on Heroku. So we need to enable the redis integration in Datadog by adding a file at datadog/conf.d/redisdb.yaml with the following content:

init_config:

instances:
  # populated by prerun.sh

Then we need to modify the datadog/prerun.sh script to add the redis instances to the redisdb.yaml file. The script is executed before the Datadog agent is started. The script looks like this:

redis_regex='(\w+)_URL=redis:\/\/([^:]*):([^@]+)@([^:]+):([0-9]*)$'
env | grep "REDIS.*_URL=" | while read -r line ; do
  if [[ $line =~ $redis_regex ]]; then
    {
      echo "  - host: ${BASH_REMATCH[4]}";
      echo "    port: ${BASH_REMATCH[5]}";
      echo "    username: ${BASH_REMATCH[2]}";
      echo "    password: ${BASH_REMATCH[3]}";
      echo "    tags:";
      echo "      - name:${BASH_REMATCH[1]}";
    } >> "$DD_CONF_DIR/conf.d/redisdb.d/conf.yaml"
  fi
done

It will fetch all REDIS_*_URL environment variables and add them to the redisdb.yaml file. The tags section is optional and can be used to add tags to the redis instance. In our case we add a tag with the name of the redis instance.

Have some thoughts? Send me an email.

After spending a couple of hours to find a new domain for the Mastodon server I finally found one that was available and that I liked: meet tacocat.space.

Feel free to follow me on Mastodon at @jonas@tacocat.space!

Have some thoughts? Send me an email.

Input boolean helpers

Setup a boolean input helper in Home assistant for every member of your household that has a iPhone:

input_boolean:
  jonas_present:
    name: Jonas
    icon: mdi:account
  eva_present:
    name: Eva
    icon: mdi:account

And here is how it looks in Home assistant:

Home (Homekit)

First, make sure you have enable and setup the Homekit integration in Home assistant. You then need to setup some automations. One that turns the helper off when the person leaves and one that turns the helper on when they arrive home. Repeat that for every person in the household.

Device tracker and Persons

By now we have a working solution that we can use in our automations. But i'm not quite happy just yet, I want to connect these input helpers as device trackers and add them to my persons in Home assistant.

You can create an automation in Home assistant that maps the input helpers state to device trackers by using the device trackers see service. That service will update or create a device tracker with the specified name. Here's my automation that create a device_tracker.homekit_* for every member of the household.

- alias: Homekit to device tracker
  description: ''
  trigger:
  - entity_id: input_boolean.eva_present
    platform: state
  - entity_id: input_boolean.jonas_present
    platform: state
  condition: []
  action:
  - data_template:
      dev_id: homekit_{{ trigger.to_state.name }}
      location_name: '{{ ''home'' if trigger.to_state.state == ''on'' else ''not_home''
        }}'
    service: device_tracker.see
  mode: parallel
  max: 10

You can now assign the new device tracker to the corresponding person in home assistant. First you have to make sure that you have triggered the input boolean once first to force the automation to create the new device trackers. You can toggle the inputs in the Home app on your iPhone/Mac instead of leaving your house for real.

Here's the final result on my lovelace dashboard:

Have some thoughts? Send me an email.

TabNine

Mindblowing autocompletion for all languages. It uses machine learning to predict what you're typing, and it's correct almost every time.

Unstable Ember Language Server

A must-have when working with Ember.js, which I do for a living at Teamtailor.

change-case

Quickly toggle between camelCase and snake_case. Or between over ten other cases.

Transformer

I use the unique lines feature quite often.

vscode-run-rspec-file

Quickly run a single rspec file or even a single rspec line from within VS Code.

Have some thoughts? Send me an email.

When I was using Jekyll I just dumped my photos cache inside the node_modules directory and I thought that I just could do that now too. But it felt dirty. And hacky.

So I search the Netlify documentation but couldn't find anything about adding files to a cache between builds until I stumble upon a comment on StackOverflow (or was it on a Github issue?) that mentioned an undocumented cache folder and that someone already had built a Gatsby plugin that simply caches the Gatsby output in that Netlify folder. A piece of cake!

1. Install the addon by running: yarn add gatsby-plugin-netlify-cache
2. Add 'gatsby-plugin-netlify-cache' to the plugins list in your gatsby-config.js file.

Have some thoughts? Send me an email.

Continious deployment with CircleCI

My previous setup allowed me to just push my updates to the Github repository and have Heroku pull the changes and update the site automatically. The new setup uses CircleCI to accomplish the same. Add your Divshot token to the environment variables on CircleCi (more info) and then tell it to deploy on successful deploy by adding this to the circle.yml file in the root of the repository:

test:
  override:
    - bundle exec jekyll build
deployment:
  production:
    branch: master
    commands:
      - npm install -g divshot-cli
      - divshot push production —token $DIVSHOT_TOKEN
  staging:
    branch: /(^(?!master$).*$)/
    commands:
      - npm install -g divshot-cli
      - divshot push staging —token $DIVSHOT_TOKEN

It will deploy the master branch to the production environment and all other branches to the staging environment. All automatically when pushing to Github.

Have some thoughts? Send me an email.

Just log in to your Heroku dashboard to enable automatic deploys to make Heroku build and deploy all pushes to the specified branch.

Read more about the new Github integration in Heroku’s Dev Center documentation.

Have some thoughts? Send me an email.

I also added a Heroku Button. So you can deploy your own version of the Jekyll site to a new app on your Heroku account with just a single click on a button. You can try it directly here on this blog by clicking the button below.

Pressing this button will deploy a Unicorn powered Jekyll site to Heroku. The blog will be blazingly fast and handle a heavy load by using Unicorn with multiple workers on per dyno. And by only using a Dyno the hosting will be completely free!

Have some thoughts? Send me an email.

It resulted in a simple planning poker app to help with the sprint planning for my remote team. I've uploaded the source to Github.

Have some thoughts? Send me an email.

Have some thoughts? Send me an email.

Every Monday morning at 8:00 breakfast is served for the Mynewsdesk employees at our headquarter in Stockholm. It became some sort of tradition for me to take a photograph of my breakfast every morning and here’s the result from 2012’s breakfasts.

Have some thoughts? Send me an email.

Gemfile

Add the Pygmentize gem to your gemfile:

gem 'pygmentize'

And run bundle install

_plugins/highlight.rb

Create a new file in the _plugins directory with the name highlight.rb. Paste the following code to make the built in code highlighter use the new Pygmentize gem:

require 'pygmentize'
class Jekyll::HighlightBlock < Liquid::Block
  def render_pygments(context, code)
    @options[:encoding] = 'utf-8'
    output = add_code_tags(
      Pygmentize.process(code, @lang),
      @lang
    )

    output = context["pygments_prefix"] + output if context["pygments_prefix"]
    output = output + context["pygments_suffix"] if context["pygments_suffix"]
    output
  end
end

Use

Just use the standard highlight blocks as usual:

``` ruby gem 'pygmentize' ```

Have some thoughts? Send me an email.

It creates a new file in the _posts directory with today's date, parses the parameters to command as the post's title and adds that as a slug to the new file. It then writes a default yaml template to the file (as specified in the script).

Running thor jekyll:new New and shiny post will for example create the file _posts/2012-12-28-new-and-shiny-post.md, populate it with an yaml template and finally open the file in my favorite editor.

How to

Add the following to your Gemfile:

gem 'thor'
gem 'stringex'

Run bundle install and create a jekyll.thor file with the following contents:

require "stringex"
class Jekyll < Thor
  desc "new", "create a new post"
  method_option :editor, :default => "subl"
  def new(*title)
    title = title.join(" ")
    date = Time.now.strftime('%Y-%m-%d')
    filename = "_posts/#{date}-#{title.to_url}.md"

    if File.exist?(filename)
      abort("#{filename} already exists!")
    end

    puts "Creating new post: #{filename}"
    open(filename, 'w') do |post|
      post.puts "---"
      post.puts "templateKey: blog-post"
      post.puts "title: \"#{title}\""
      post.puts "tags:"
      post.puts " -"
      post.puts "---"
    end

    system(options[:editor], filename)
  end
end

Use the new command:

$ thor jekyll:new The title of the new post

You can even specify which editor to open the new file with:

$ thor jekyll:new The title of the new post --editor=vim

The default editor is Sublime Text 2, just change on line 4 in jekyll.thor if an other editor is preferred.

Have some thoughts? Send me an email.

Having fun with Loopcam.

Have some thoughts? Send me an email.

This work really out of the box if you host your site on Github via Github pages but if you, like me, are using Heroku you need to set up some more automation. See my previous post on how to automatically deploy to Heroku when pushing to Github.

Have some thoughts? Send me an email.

I found this little Sinatra app a couple of days ago and have now set up it to automatically deploy this Jekyll site to Heroku when I push the repository to my Github account. It receives a post from Github via the post-receive service hook, fetches the updates and push them to Heroku.

I have made some changes to the app, so use my fork until the pull requests has been merged. My fork adds support for private repositories and makes it run smoothly on just one free Heroku Dyno.

Just follow the instructions in the README and you're done in less than 5 minutes.

Update 2014-06-10

I've have switch to a more competent solution for my continuous deployment to Heroku.

Update 2015-02-01

Heroku has released their own Github integration.

Have some thoughts? Send me an email.

Rack-Jekyll

I have previously used jekyll --serverin my Procfile on Heroku which in turn generated the jekyll site and booted up a webrick server. But I wanted to run my site with Unicorn and it's support for multiple workers on a single Dyno. And to do so my Jekyll site must be a Rack app.

Rack-Jekyll to the rescue! It was exactly what I was looking for, it turned a Jekyll site into a Rack app. I had some issues with it on Heroku and made a pull request with the fixes. At this moment it has been merge with the master but there is not a new version released of the gem.

Jekyll + Unicorn + Heroku

Jekyll-Heroku-Unicorn is a sample app of a simple Jykyll site with 6 Unicorn workers ready to be deployed to a free Heroku Dyno:

1. git clone git@github.com:himynameisjonas/jekyll-heroku-unicorn.git
2. cd jekyll-heroku-unicorn and run bundle install
3. heroku create followed of git push heroku master
4. heroku open and you're done!

Have some thoughts? Send me an email.

If you have a Gemfile, start by adding the fleakr gem to your Gemfile and run bundle install. If you don't have a Gemfile, install the gem with gem install fleakr.

source :rubygems

gem 'RedCloth'
gem 'jekyll'
gem 'fleakr'

Then add the flickr plugin to your _plugins folder:

require 'liquid'
require 'fleakr'

Fleakr.api_key       = "XXX"
Fleakr.shared_secret = "YYY"
Fleakr.auth_token    = "ZZZ"

CACHED_IMAGES = {}

module Flickr
  def flickr_image(url)
    "<img alt='#{image_object(url).title}' src='#{image_object(url).large.url}'>"
  end

  def flickr_medium_image(url)
    "<img alt='#{image_object(url).title}' src='#{image_object(url).medium.url}'>"
  end

  private

  def image_object(url)
    CACHED_IMAGES[url] ||= Fleakr.resource_from_url(url)
  end
end

Liquid::Template.register_filter(Flickr)

Read the readme for the fleakr gem on how to obtain your flickr api tokens. Then add them to the top of the flickr.rb file.

Loop over your photos in your post layout, for example like this:

<article>
  <h1>{{ post.title }}</h1>

  {% for flickr_url in post.flickr %}
    <a href="{{ flickr_url }}">{{ flickr_url | flickr_image }}</a>
  {% endfor %}

  {{ post.content }}

  <footer>
    {{ post.date || date:"%Y-%m-%d" }}</br>
  </footer>
</article>

Finally, the last step is to add one or more photos to your post:

---
title: A blog post with photos from flickr
photos:
  - 'https://photo-storage.brusman.se/6167152072.jpg'
  - 'https://photo-storage.brusman.se/6167152448.jpg'
  - 'https://photo-storage.brusman.se/6167153574.jpg'
---

Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor
incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis
nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat.

I have also published all these examples as a gist.

Have some thoughts? Send me an email.

And this is the awesome thing that happens when one of the posts of the same site as above is posted and gets popular on reddit. From a stable 150-200 visitors per to a huge peak at 20 000. Such a big difference that the rest of the days looks all flat.

Have some thoughts? Send me an email.

However, the plan is to post some photos and maybe even some things about my other hobby (besides photography) coding/web development/yadayada. Time will tell.

Have some thoughts? Send me an email.