Seeing the OG images I generate for this blog has me thinking. If you split a blog post by paragraphs and generated PNGs, and then you stitched those into a GIF with the frame time based on text length... that would maybe be a portable / embeddable blog post? Not good for accessibility and not to replace the post, but maybe an interesting derivative.

This site is a bash build script that concats HTML together. I also put together a little start script using fzf that has some common task options new_post / build / commit. I really like the feel of the starting bash script - it makes it feel like writing in this blog is its own little dedicated computer program.

The source for this one is viewable on github.

I think I'm going to try a new blog concept (a continuation of the thread-focused blog idea), that combines what I do on feed.

Thinking about things that make you see the world around you differently: ARGs, birding, mushroom hunting, simulating physical phenomena in code, drawing.

I was listening to the Egglant podcast interview with the Animal Well maker and they mentioned "secrets in plain sight" in the game and compared it to ARGs, where your street becomes something new/different to you because there is a geocache (or pokemon station there).

That made me think about a tweet I saw yesterday from Jer Thorp about how birding helps you to see more details around you. Someone else mentioned taking up mushroom hunting had a similar effect.

What if I made a blog like this blog but also more like Twitter. Where I could make little posts, or I could also reply to old posts, bumping that 'thread' of posts to the top. But (like Twitter) it would show the update, with an indication and ability to click through to see the full thread.

That seems kind of cool? Like it'd be good for long-running thoughts I keep returning to and also for WIP posts...

Maybe structuring stuff in the code gets weird, but it's going to be small-scale so I think I could figure something out. I think I'd just do markdown files with like a link at the top, and use the filename for the date slug like I've been doing. I would still like titles I think. But maybe that's just otional.

I would say this garden has been a relative success in terms of helping me write more. I kind of want to combine it with my feed, and maybe my homepage?

Feed would be the thing to do first I think. It's interesting trying to categorize the differences between posts. Garden is more writing based. Feed is structured around an image. Feed was designed to make sure I wasn't blocked trying to think of a title, that I could do more work-in-progress stuff. Maybe that's different now that garden is here.

Overall I want to keep things simple. But I also want to have some fun. I'm pretty confindent I can figure out ways to do what I want, and as long as I keep the file format relatively stable, markdown for garden and a common structure for feed, I think I can do it without really risking anything.

One of the best things I had going with my old linux set up was my gif recording workflow. I had a few helper scripts connected to dmenu (and then I think rofi and fzf), where I could record the screen and then convert into gif. It was still a bit janky. But it made it pretty quick to take work-in-progress clips and post to my feed.

Since I tore down my set-up and rebuilt on Ubuntu server install and Nix home-manager I haven't had a good gif set up. Tonight I think I got it fixed up though I'm sure I will continue to tinker.

The bulk of what makes it work is in gif.sh. The majority of which is from this comment by IntelligentPerson_ on Reddit. Slightly modified to fit with keyboard shortcuts. It's even got nice little notifications.

Through work, I've been playing with the Gemini model's ability to return bounding boxes for an image. I think I want to try and use it to make some sort of image (sketch or photo) to website generator. Maybe leaning into skeuommorphism of like a photo of a physical desktop that contains all the links to my projects as items on the desk.

A very sketchy proof of concept

Here's a brainstorm:

Through work, I've been playing with the Gemini model's ability to return bounding boxes for an image. I think I want to try and use it to make some sort of image (sketch or photo) to website generator. Maybe leaning into skeuommorphism of like a photo of a physical desktop that contains all the links to my projects as items on the desk.

A very sketchy proof of concept

Here's a brainstorm:

I'm not sure if it was from reflashing the keyboard firmware or resetting the computer software but it's working now. It is nice. I'm not sure what lesson to take from this.

Am I writing too many posts now? Still getting a feel for this thing. It is fun the ways different writing environments seep into the writing.

I'm having trouble pairing my wireless keyboard with my mini PC. I keep trying new things and searching the proglem in different combos.

My current theory is maybe the bluetooth driver isn't up to date enough? But I think it actually may be.

The haunting thing is maybe it's just a bad state issue, and if I reset the right thing (firmware on the keyboard, bluetooth status on the keyboard) it will be all fixed up.

Trying to keep up the habit of quick writing in here. Now morning writing from the couch, the dog beside me, the baby exploring a cardboard box. A nice spring-like morning.

Did some organizing syncing the setups yesterday. Still liking the Ubuntu + Nix configuration. Getting better and faster about keeping it up to date across the laptop and the mini PC. The server's a bit behind.

For the glassese set-up I got audio and bluetooth working yesterday, so I think everything I wanted to do with that thing is proved out. I ordered a cross-body bag that I hope will fit everything well. Porch use has been good so far, interested in trying a commute with it. The battery pack has been working well, I'm surprised how much having an exact percent readout, rather than something like 4 lights, helps me feel secure about power draw.

Hopefully the first of many posts written while sitting on the porch. This is the first one I've got set up with the AR glasses.

It was cool earlier when the light meant I could see more of the background, so it felt like a floating transparent screen. Now it's darker and the screen is less transparent.

I'll do a bigger post on the set up later, but I can list out briefly what I have going:

This blog!: I want to get preview images working. I'm trying to generate them using cairo, so I can do it as part of the build script. I would say I've proven out the basic concept, now need to decide on some text positioning and hook it all together for the first run.

Mini PC: I'm working on getting a mini PC set up with the my current go-to Ubuntu server + Nix home-manager config. Eventually I want to try and make it a wearable deck to go with my AR glasses. Will write about that more later.

I want to make sure I can write from anywhere, including my work mac, which I'm writing on now. I had to update bash on the mac (I did it using Nix home manager). Otherwise things seem to be working well.

Edit: also had to replace a grep P with a sed.

Today I made this blog / garden. I want it to be a place where I can write quickly and with focus.

Right now it's a bunch of bash scripts that put together some templates. I wanted to know exactly what was going on so that I'll be able to modify it in the future. Though to be transparent the bash scripts were AI assisted. Still, I think I understand all the steps.

I want to keep things short.

Something I think a lot about is what level of abstraction to attack a problem on. Here is a loose cataglog of where I've been thinking about it lately.

React libraries

Previously I've been reluctant to use libraries in my react code, generally preferring to roll my own solutions, believing that you tend to hit a wall with what you can do with libraries and working around their limitations becomes as much or more work than building your own. There are things I would use, like state management libraries, but I figured most of the interaction stuff I'd rather do myself.

Rough draft of my current tools-for-thought thinking.

I'm excited for all the energy around tools for thought. Some notes on my current thoughts (probably to be revised in future entries).

Blocks as the basic unit

Blocks as a basic, contained, and usually addressable unit feels to me "right". You can see it in Notion and Roam and many others.

I'm excited for all the energy around tools for thought. Some notes on my current thoughts (probably to be revised in future entries).

Blocks as the basic unit

Blocks as a basic, contained, and usually addressable unit feels to me "right". You can see it in Notion and Roam and many others.

How do you change things?

I'm interested in users having the ability to personalize websites. I've mainly thought about it as a case for UI that lets you choose different options, probably powered by CSS (see components.ai's CSS.GUI).

My CSS Paint project lets you use a brush. It's interesting to use a brush. Should we let people style their website using a brush to paint styles on to divs?