Confidence

This is one of those more personal blogs.

This week, well, started last week, I had a small technical challenge. The exact details do not really matter for this. But I'll explain.

I wanted to code something - something that was not quite readily available (some people had things very close to something I could just use, but not quite, which is where it gets fun).

The boring bit: It was a system to decode PNG image files, and allow me to use them in an embedded system driving a e-paper display. The details do not matter.

All that matters is that it was not easy to do - it involved understanding a detailed technical specification (PNG), and a new thing (zlib) which I had not used before. For a change the specification is a really really good one, which, in many ways, makes it easier.

Can I do it at all?

A biggie for me is should I, can I, even embark on this project.

To be clear, I have no reason to. Nobody wants this (well they may now), nobody needs this, nobody is asking for this, nobody is paying me for this. It is solely to make an e-paper thing I have easier to use. It is educational for me, a challenge.

Indeed, many of the helpful suggestions are to use a different file format, and not png. I do that already, but I wanted to make it simple - all can handle "save as png" I feel, so can I make it handle png. All and every format for png though?!

The protocol - the file format, which I literally sat in hot tub for over two hours reading, is good. But has some tricky bits. One is the compression/expansion using zlib, something I have never done.

I was seriously concerned I could make this work at all. Could I do it? Could I do it in a way that fitted in memory on an ESP32? Am I up to the job?

Yes I can!

I decided to go for it, in linux and then ESP32 code.

Each step worked well, and was not an issue. Minor changes as I coded. Lots of testing. zlib is easy, it turns out. zlib works on ESP32 too.

PNG has a lot of options, and as a decoder you have to handle them all. As an encoder you can be picky, but decoders cannot. I had to slog over the different options - make every one work properly.

In a day?

I spent a day, well, until early afternoon, and then a couple of hours next day. Total time under one day's work.

It works!

It went perfectly - the bugs were simple to fix. When I slept on it, all the remaining bugs came to me in my sleep, like they used to.

Not lost it?

So no, I have not lost it!

The work was not as easy as it used to be - younger me would have worked on in that evening and not next day. Younger me would not have made as many typos (way more since I had a stroke). But the results is good.

I worry about "losing it", and "imposter syndrome" all the time. I hope this is is a good sign I am not a losing the plot, honest.

Here it is https://github.com/revk/ESP32-LWPNG

Solar System goes live

As I have said, I have been working on a number of boards as part of a project that provides access control and alarm functions in a modular way: https://github.com/revk/SolarSystem

I have test systems on my bench, and I now have a small system at home. It is all working well and has helped me iron out some of the bugs.

But this week is the first proper system with 28 live WiFi nodes, meshed, and linked to the back-end cloud control. Scary stuff. It has bell boxes, keypads, PIRs, reed switches on doors, fire alarm inputs, even a panic button in a disabled toilet. Importantly it has a lot of doors. The design is pretty robust, and the whole project is all open source.

My case is packed, and boxes of tools and parts are all ready. It was a lot of work making all the modules. A lot of time with a steady hand with tweezers. I even have half a dozen spares, just in case.

So the night before I head off to start the install, I have imposter syndrome kicking in. How did I think I could possibly design and make a complete access control and alarm system from scratch (PCBs and s/w)? Well, seriously, I need to give myself a kick - I have been doing this shit long enough to know this is bullshit. It will be fine.

Assuming all is well (and I know there will be teething problems, bugs, and features, all of which will need addressing), my next big challenge is whether I can progress this in to a proper product to sell. At the very least, to make the modules (as pictured above) something we can legally sell.

In the mean time, other hackspaces that are interested, do get in touch, and I can help you set up such a system.