Got an app idea or need some coding done? I can help! Learn more about how you & I can work together.

Wintermute’s the Name, JARVIS’s the Game

An announcement post about my growing side project, Wintermute!

:book: wintermute :bookmark: personal-goals, jarvis, iron-man, wintermute, projects :clock7: :clock3: about 6 minutes

When people ask me what I do, I almost always reply with the sentence, “I’m a full stack web developer”. It’s not a lie, but it’s only what I do because it helps me pay the bills. That might read wrong to some people, but web development has taught me a lot about remote system management and database management, so it has an immense value add to my end goal.

Before continuing, I strongly recommend becoming acquainted with JARVIS.

Tony shopping on Amazon.

Jarvis is practically and not yet entirely what I want to build out of an self replicating operating system. That was the biggest aspect of the platform; it has to be able to spread itself whenever necessary. In conjunction to that, it has to understand how far it spread out. It needs to know its environment and what it’s capable of doing in said environment. It also needs a level of grasping context. I originally called it Taylor, but after a recommendation from a friend; the name’s Wintermute.

Run, Forest, Run

The one thing about computers that make them so useful to us is their abundance. One computer on the planet wouldn’t be able to hold Facebook, Google, and Microsoft. But farms and farms of them can and have them run insanely fast. There’s all of that good and juicy stuff like remote procedure calling, shared memory and core processing clustering. All of that are things I’d want to incorporate into the foundation of this “big idea” of mine.

A Internet of Things

Another reason for multiple computing units is also for I/O. The platform should be able to receive data from a multitude of locations and (re)distribute it across those same devices. Think of this as a way of creating many faces for one person. The simplest point of entry would be a microphone and speaker system. Both devices would be connected to a small computing unit whose sole purposes is to directly route that captured audio to a capable machine for parsing said raw audio and a device that can play back said device without much difficulty. That could be considered the simplest form of a human interfacing unit. An extended version would allow for a RCA jack to permit more capable devices to extend the functionality of the device. A key aspect of the physical (and digital) components of this project is that everything is extensible. Extensibility can ensure greater form and function, in my opinion; correct me if I’m wrong.

Understanding its World

The following text is going to be heavily opinionated. And it’s on the (little) research that I’ve done about knowledge representation and interpretation in computing.

There’s three aspects of the building blocks of usable attributes we collect from our environment. They’re in the form of data. Data, when joined with other bits of data, becomes information, and when information is used in a specific context, it becomes knowledge. My (extremely debatable) definition of each word is as follows:

Building Wintermute

Right now, I’m rewriting the basis of the Wintermute platform. It’s a bit weird how I have it envisioned but it’s something like this. Wintermute would start up in the background, right before the desktop shell is loaded or the login manager, probably even before X is started. For now, it’s started from the console within a X session. Like a typical extensible application, it searches for a core instance lurking about and starts one if none exists. It loads extensions and then falls to the background, only acting when requested.

Now, the function of it really exists within the plug ins. The way I saw it, Wintermute has three extensions that are vital for its operation, the DataApi, NetworkApi and LinguisticsApi. From the names of those libraries, a bit can be derived.

Data in Wintermute

One of the biggest plagues in software applications are object document mappers (or relational mappers, if that’s your fancy). What I’m aiming to do is store data abstractly. It’s going to cause a tiny overhead problem since everything will need a middle format, but it’d allow for a plethora of data storage solutions like Mongo, MariaDB/MySQL, or even a REST API. It’s still in the flux of development.

Networking with Wintermute

Now this is the fun part. One theme of Wintermute is to make protocols and barriers as invisible as possible. Data and communication should be able to occur everywhere. So with networking, I initially had the idea of having listening sockets on every kind of socket (ICMP, UDP, TCP or ARP). Obviously, this wasn’t too much of a good idea due to my lack of knowledge of how to tame this things and since connections can be abstracted away only so far. For now, my solution to this might be implementing a central hub that represents a central stream of data and have workers that connect and translate data accordingly. I figured this to be way more efficient so that if data drops on UDP, we could attempt to reestablish a connection via TCP. I also saw it as a way to introduce more native communication protocols like serial communication and Bluetooth. Awesome!


Now, this is my bread and butter of Wintermute. Probably the most experimental part of the entire application, I’d attempt to build a hot swappable semantically accurate linguistic engine using the theory of link grammar. The only issue with this is that some languages like Farsi won’t work with it. Since I don’t know it, I’m not stressing it too much, but I do hope that someone would come along and help with it.

That’s my idea. What do you think? Lemme know!