Visual Studio Code is a nifty all-in-one editor/debugger for AutoHotKey, but the language hasn't defined cold-folding regions like other languages such as C# (#region/#endregion), or JavaScript (//#region///#endregion).

I found a good active plugin with syntax-highlighting for AutoHotKey's .ahk files, and got to work with a proposal to implement folding:

1. I double-checked folding conventions for other languages.
2. Learned how VS-Code Language-Extenstension features are defined with examples.
3. Checked my usual Regex resources and examples for guidance.
4. Tested my concept locally in VS Code.
5. Started discussions for the Enhancement Request in GitHub.
6. Created a Pull-Request with my code using GitHub's new propose a change feature.

The function was written as a work-around for a client with O365 which we didn't have administrator access to. Before this little tool: it was considered normal to bounce a ticket to vendor (3-day turnaround) just to get a login-name, or to diagnose why a caller's Skype was no longer connecting. I (ab)used the old V1 Microsoft Graph API module to scrape the client's full graph for login names, SIP, aliases, staff id, and licenses... then cobble in some less ambiguous product names, and spit out a tab-delimited .csv file.

This is a stripped out example of tool-code that was actually used in production quite frequently. It was hacky and quick, and only a POC which unfortunately never got approved for development... so instead, I slapped some control-code within; at-least my colleagues could also use what we had on-the-fly. Time savings from this script alone gained me about 5-10 hours per week, with ~5 other techs using it daily, we'd consistently save 20+ hours weekly using this simple script.

With more time I would have liked to:

• Add some nice dedicated controller scripts and keep the module pure.
• Create module manifests and tests.
• Update to support PowerShell core / v7.
• Code in a fallback to the real productNames (for any license names I hadn't overridden).
• Break those advanced-properties into a second step so the lookups (licences, aliases etc) don't holdup the core loop. 3x lookups on 200k users not using parallel pipelines becomes slooow...

This Python 3 RESTful* server is what I used as a simple queue service while I was testing my game server mod. The videogame only supported HTTP GET and POST (rather than PUT/PATCH) so it would respond appropriately to those requests.

I programmed some buttons on my StreamDeck device to send requests to this API and tested it live with about 8 friends. Later I added some "quick command" endpoints which my friends could trigger (while I was away) by simply refreshing pre-defined webpages.
This is why the app went from being a pure REST server, to just RESTful functions.

The front-end was beyond the scope of this example hence choosing Bottle as the lightest, fastest option for prototype testing. I would consider Django (with authentication, templates, and a persistent database) to support a production front-end.

Screeps is an online game where you battle your army of bots on (what they call) a real-time strategy battlefield. It's really turn-based gameplay: Every second-or-so it will run all your JavaScripts top-to-bottom, serialise it, execute on the C++ game server, and then deserialise the whole game-state as a JSON string to run from scratch again next tick.
As such, I don't think callbacks etc would have any real use to a simple bot. I fluffed-out the Screeps tutorial by breaking it out into more modular functions... functional is definitely the key-word here.

Once I get my head around game ai StateGraphs I am interested to rebuild this as an object-based bot and then compare performance.

My fascination for automation got me thinking about general workflow shortcuts and GUIs which we have less control over. I have seen AutoHotKey thrown around before and found a good opportunity to give it a whirl.

If I didn't quite catch something while watching YouTube on another monitor, I can't just use my media buttons to jump back 10 seconds to repeat that last bit. Instead I break my focus, click on the YouTube screen (which also pauses it), tap j to scrub back, resume... and finally go back to what I was doing.
After looking into ways to send FF/REW from my StreamDeck or remapping my keyboard's media keys, I found issues with the way background apps (Spotify, MediaPlayer...) occasionally take focus of those commands over Chrome+YouTube.

If I wanted to control YouTube directly, Chrome has to be launched in developer mode - which limits the possibility of ad-hoc use. Not-to-mention that I use multiple browsers at any one time.

An easy solution right in front of me was to write an AutoHotKey which hooks on the media keys, and sends the commands directly to the last known YouTube window...