@gavhu10

I fixed another bug, learned a new thing about C, and wrote some docs. When there was a space on a line by itself the device performed as expected. But when something that was not a function or something on a line by itself, it typed the first two letters, then inserted a newline. After hitting my head against walls, I found out that I had been using the strcomp function (in C) wrong. It returns 0 when a string matches and I expected to return 1 instead. That caused all the lines to be handled like they were just a space. I also found out that I had not accounted for a newline character and had assumed it would be removed sometime in the parsing process. I managed to figure it out, though, and it works fine now. I also wrote some docs. I wrote them in markdown and made a converter that turns them into html and stores them in the device when the project is compiled. That way someone can read the docs online or from the access point.

I added a bunch of QOL improvements. I made it so that Ctrl+S would save the file in the editor. I made it so that a profile is automatically selected when the web ui loads. I also added mDNS so that you do not have to remember an ip address but can just go to wifi-duck.local!

I fixed a tricky bug. If the script parser encountered a line with just a space, it would not delete the line like it would if the line was empty. Instead it would split the line on that space and return a NULL pointer. The reason this was hard to find was because it was deep in the internals of the parser and was written in C and I could not use print statements as the Serial Arduino is C++ only. When I could use print statements, I did, and with careless abandon! Then I spent a bunch of time making my commit message line up nicely.

I made it so that the selected profile and selected script are highlighted. It took longer than it should have because I have not learned frontend and do not use AI. Another problem is the time it takes to upload changes to the device. However, I still managed to make it look okay.

I added media keys! So far I have pause/play, next, previous, and volume. It turns out those types of keys are not regular keyboard keys. They actually are part of something called Consumer Control. The library that works with Consumer Control is very similar to the built-in esp-idf keyboard library but as I looked at the code, I noticed something wrong with it. I found out that the USBHIDKeyboard library was vendored and then modified. To make sure that Consumer Control did not mess with the keyboard I refactored the code so it just used the built in library. Here is a demo of the volume keys:

I rebuilt the case! It took ~20 iterations of export, slice, and print, not including failed prints. It now covers the whole top of the board but includes some vents so that you can press the reset and boot buttons. It took a lot of work and I changed the design multiple times. I ended up with hexagon holes in the case which in themselves were tricky. I used an OpenScad library I found on thingiverse and modified it, only to redesign the pattern so that those edits were not needed. Typical! The main picture is what I finished with. I also have pictures of the other designs.

I made alternate keycaps with OpenScad for my keypad but in the end I chose to stick with what I had before. I wanted the buttons to have smaller profiles and be the same colour. At first I tried to make a complicated clip like the injection molded ones that came with the set, but that failed as 3d printed material is not suited for that purpose. Instead I made keycaps that held on with friction. That worked, but they were hard to take off again and they ended up being uglier than the injection molded ones.

I made a case for my Keypad using OpenScad. It took eight iterations of export, slice, and print, but I like how it turned out. It makes it look a lot more finished. I tracked the time with Lapse because I could not track it with hackatime.

I added a command (function?) called STRINGLN. This is just like the STRING function, which types out a string, but after typing out the string, it presses enter. This is not a big change, but to make it, I had to read all the code involved with parsing the script and pressing the buttons. STRINGLN is also a very handy function (command?) as it reduces verbosity and many existing duckyscripts use it. I created a pr to the upstream project which you can see here.
  
Below is an example script from the Hak5 repo that uses STRINGLN.

I integrated the NeoPixel LED with the firmware. My only problem was that the NeoPixel library was vendored in. I fixed that by adding a PlatformIO library in the config. Now you can show the status of your script through the LED! It also shows you if it managed to connect to WiFi. I tried to make the ip address of the device static while in station mode, but it turns out things like that vary between routers, so I just added a readout of the ip address to the settings. I made a demo video for the LED. The three flashes at the beginning confirm that it connected to a wifi network. I had to put tape over the LED to diffuse it.

I made it so that you did not need to connect to the access point to configure the device. This is handy because otherwise you would have to disconnect from the internet to configure the device. I also fixed a stubborn bug with the profiles. The problem was that the get_profile command returned the current profile with a newline. This in turn broke the javascript that let you edit the scripts. Instead of editing 6 (for button six) it was editing \n\r6. This broke the quotes and made a big mess.

This is a really long devlog because profiles are a big feature. They make it so that you do not have to overwrite your macros every time you switch contexts. They were tricky to make for a variety of reasons. I had to add multiple cli commands so that the web ui could do things, refactor lots of things so that they could be reused, and change how I compiled the project. I also had to write frontend code! shudder I learnt a lot of things about the project while debugging. For example, a cli command called from the websocket must return something or the ui will hang. Fun times! Here is the diff of what I did: link

I made so that the ordering made more sense. It used to start with zero which is fine but confusing for some people. I also made so that the pins are read in the right order, as the order used to be

1 4 7
2 5 8
3 6 9

to make it easier to route, but that also is confusing, so now it reflects the ui. The next step is shortcut profiles like the og duckyPad.

I made it so that the script selector is like the buttons so that it is easier to use. It took a while because I am not very good at frontend things, but I like how it turned out. It was made especially tricky as on-device debugging involves compiling and uploading the new firmware, connecting to the access point network, and then reloading. I did get a debug setup going with python -m http.server but then the websockets did not work for obvious reasons.

This devlog is rather long, but a lot of things needed to happen before I wanted to make another devlog. I had a very difficult bug involving macros (written in DuckyScript) running at the same time. It wasn’t always a problem, but the buttons made it really easy to crash the system by overlapping scripts. When I added some code to keep track of that, it would crash when launching a script from the web-ui, but not when launching them from the terminal or buttons. After much (much much much) debugging and banging my head against walls, I figured that the problem was with a websocket status pulse. By delaying the pulse, I could avoid a crash. You can see the one line javascript patch below. I’m not sure why it was crashing the device, but it seems to be fixed now. I made a PR to upstream here Anyway, enjoy the demo!

I made a macropad with KiCad. It uses an esp32 s3, so it will have a web based config editor. This project is the firmware for it as I did not know that stardance could do hardware projects. Since I just worked on some button code and got set up, I will just have some pics of the macropad.