One possible future for programs will be a solid core, with a plugin architecture that can hook almost anywhere (think WordPress or the old XUL based Firefox) and is capable of protecting the program well enough.

On top of that people can vibe-code their plugins. The conversations that built them would be stored, with compacted contexts and plans. On core update or new extension addition everything will be regenerated.

Sharing would be done only through extraction of specs. No code would be able to be shared because unless all of the plugins are generated at the same time, there is no guarantee that the code would work.¹

Everything would become slightly less stable, but the base programs could become simpler. A cottage industry of maintainers and customizers would emerge.

Also see Warranty Void If Regenerated.

I want to keep a journal, I have three requirements:

1. The data must be stored as plain text and plain images on disk, that I own.
2. The entries must be labelled with their creation timestamp, location, and ideally weather information.
3. I want to be able to create a journal entry from an image. The above metadata should be loaded from the image.

I started journaling with Day One¹. After it got sold I have lost faith in its future. They also use a proprietary database with limited export options which fails my condition 1.

Apple’s Journal looked interesting, but it does not satisfy my conditions 1 and 3.

I looked into self-hosted options:

• Journiv² — has a terrible frontend, uses a database instead of markdown, has asinine features like “reflect now” that can not be turned off.
• Obsidian — would be great but does not support 3 and requires their paid sync service on iOS.

Diar - iOS shortcuts for authoring journal entries

When I squinted at my requirements I realized that what I need is a Hugo³ blog with a login and different syntax + some goodies, and a couple of iOS shortcuts.

I always start programming by defining my data structures. First, I’ve decided on the syntax of my entries:

Since the only mandatory metadata is the timestamp, the obvious format is a markdown file with the timestamp as filename and the entry in the text.

For the other metadata I wanted something human readable and editable. I didn’t want to use a preamble like Hugo does.

My solution is to add a footer using the --- markdown syntax for a horizontal rule. Then use hashtags for tags, lines starting with /images/ for images, location prefixed by @ and a simply parsed weather. If there is a line that matches neither, it is interpreted as plain text location.

E.g.: in a file named 2006-01-02T150405.md:

This is an entry inside my journal, as you can observe it just starts as a plain text.

You can use markdown. I only really use emphasis and blockquotes.

---
/images/a-nice-illustration.jpg
#rhea #journaling
Home, France
@48.8687162,2.2219584
17°C, Cloudy

All of the footer entries are optional.

It was easy to create two shortcuts in iOS to:

1. Grab the current location, weather and address and write a new note with current timestamp as filename with this information. Then open the file immediately with iA Writer.
2. Receive an image, use it’s exif to get the time and location. If the date is today, grab the weather. Copy the image into the desired location, create the note and open it in iA Writer.

The advantage of handling images this way is that iA Writer supports displaying them in preview mode.

For the first use case the shortcut looks like this:

Rhea - Go server to render the journal in a nice way

Since I knew Hugo was almost what I wanted, I decided to learn Go and write my server.

My design goes roughly this way:

1. Fully server-rendered html page.
2. No static rendering, live-running server only.
3. Single user, the login credentials are passed through environment.

The program should do the following.

1. On startup look into the directory that contains the entries.
2. Iterate over the entries.
   1. If the entry contains an image, generate a thumbnail for it if it hasn’t been done and store it in a cache directory.
   2. If the entry contains a location, use MapBox to fetch the map, and store it.
3. Start watching the entries directory and update the in-memory representation if anything is added or changed.
4. Start a web server that serves:
   1. / - At root show list of the first page of entries, show the location map if the location changed from previous entry by more than 50km. Show images as thumbnails.
   2. /after/TIMESTAMP and /before/TIMESTAMP to serve a next/previous page.
   3. /memories/TIMESTAMP - Show a detailed entry with full size images and a more detailed location map.
   4. /images, /thumbails, /tiles-ZOOM, static - Just serve the files.
   5. /login - Render this page if there is no valid session cookie set.

And that’s about it. Cook this with some nice templates and a bit of CSS and we have a journaling app.

If you would like the source code feel free to ask, I am no longer publishing my repositories publicly.

Finally some good new has come out of the AI world.

If you thought piping curl to shell was bad, the AI crowd has decided to turn the knob to 11.

Thanks to the likes of Openclaw¹, we will soon be able to hack anything and everything, the way 90’s movies convinced us we could.

And we will be able to do it basking in bespoke vibecoded UIs, as Operation Swordfish has intended.

I started lifting 10 years ago using the Starting Strength 5x5 iOS app¹. I would still recommend it for beginners.

Then when I switched programs to 6 day Push/Pull/Legs ², I switched to a paper notepad with just handwritten weights and reps like so:

Since my progression solely depended on the last time and the goal was to do “moar”, with a 6 day split it was easy, just flip 3 pages before and compare.

Eventually I started using multiple gyms, and didn’t have the notebook with me, so I switched to a long note in the Notes.app on iOS. There the comparison to previous time was more tedious, one had to scroll up and down.

Naturally I made an app. I started by making it an iOS app, but quickly abandoned that since I realized nobody will ever use it and I do not want to go through the review process.

Then I made it into a progressive web app.

You can find it here: https://apps.yozy.net/swolog/web

Note that I made this app for me. I am sharing it with the world since it is based on many open source contributions. I am licensing it under AGPLv3, but I provide no support, or would answer for any feature requests.

I bought this toothbrush almost exactly 4 years ago. It replaced my then 4 and a half years old Oral-B PRO 3000 toothbrush. Today, the Series 8s is dead, due to failed battery. This article is a little post-mortem, a review after the whole lifetime of the product.

Let us start with the good parts:

• The toothbrush cleans teeth well.
• The best feature is that the pressure sensor not only indicates when too much pressure is applied (all decent electric toothbrushes do this), but also indicates when you do not apply enough pressure.
• You can customize the color of the LED and it has RGB, which is always great.

However this is where the good parts end. The main problems with the device are:

• It is expensive, and so are the replacement brushes.
• Its smart features are pure gimmick: the position tracking does not work and nobody needs to track their brushing frequency.
• The travel pocket is way too small to actually fit the toothbrush properly.
• It gunks up easily.
• Too many brushing modes make the menu tedious to use.

The battery is so bad it merits its own paragraph. The toothbrush became unreliable after maybe two years of use. Randomly, in the morning it would drop charge from 100% to 0, suggesting faulty electronics. Even at its best it would not last through a 10 day trip.

Verdict

Do not buy this, and steer away from the iO line in general. Myself, I bought a newer model from the PRO series for two reasons:

• The battery in my old one still work and I use it when I go visit my parents.
• You can now get off-brand cheap (and ecological) brushes for the from The Humble Co.