JMJAtlanta.com

<disclaimer> These notes are my understanding by looking at code. I may be wrong. Do not take my word for it, verify it yourself. Much of what is below is my opinion. As sometimes is the case, following good practices the wrong way can make a project worse. That is not what I’m talking about here.</disclaimer>

Recent versions (0.20ish) of Bitcoin Core attempt to reorganize logic into objects with somewhat related functionality. In my opinion, this is one of the best ways to reduce complexity. External components communicate with the object with well defined interface. This can lead to decoupling, which makes unintended side effects less likely.

I am focusing on the NodeContext object, as I believe this would be a big benefit for the Komodo core code. I am using this space for some notes on how Bitcoin Core broke up the object model. I hope this post will help remove some notes scribbled on paper on my desk. And who knows, it may be useful for someone else.

Breakdown of NodeContext

The NodeContext object acts as a container for many of the components needed by the bitcoin daemon. The majority of the objects are std::unique_ptr. This is a good choice, as copies are explicit. Below are the components within NodeContext:

• addrman – Address Manager, keeps track of peer addresses
• conman – Connection Manager – handles connections, banning, whitelists
• mempool – Transaction mempool – transactions that as yet are not part of a block
• fee_estimator
• peerman – Peer manager, includes a task scheduler and processes messages
• chainman – Chain Manager, manages chain state. References ibd, snapshot, and active (which is either the ibd or the snapshot)
• banman – Ban manager
• args – Configuration arguments
• chain – The actual chain. This handles blocks, and has a reference to the mempool.
• chain_clients – clients (i.e. a wallet) that need notification when the state of the chain changes.
• wallet_client – A special client that can create wallet addresses
• scheduler – helps handles tasks that should be delayed
• rpc_interruption_point – A lambda that handles any incoming RPC calls while the daemon is in the process of shutting down.

One of the largest refactoring challenges in Komodo is the large number of global variables. Pushing them into their respective objects (especially chainmain and args objects) will help get rid of many of those globals. This will help compartmentalize code and make unit tests easier.

I began a personal project a long time ago to provide connectivity to various cryptocurrency exchanges. I never got around to finishing it. It did however stay in my mind. And while working on other projects, I gathered knowledge of better ways to implement different pieces.

Recently I began anew. The market has changed a bit, but the basics still hold true. There is a need for fast connectivity to multiple exchanges. Open-source “one size fits all” solutions that I researched did not work for me. They did not focus on speed and efficiency.

The Problems

Centralized exchanges are still the norm, although distributed exchanges (i.e. Uniswap) are making headway. While I believe distributed exchanges will be tough to beat in the long term, currently their centralized cousins provide the “on ramps” for the majority of the casual users. They also suffer from a dependence on their underlying host (i.e. Ethereum in the case of Uniswap). These are not impossible hurdles to overcome, but for now centralized exchanges play a large role in the industry.

Regardless of the centralized/decentralized issue, there are multiple exchanges. Those with connectivity to many of them can seek out the best deals, as well as take advantage of arbitrage opportunities. But this is a crowded field. Getting your order in before someone else becomes more important as bigger players squeeze the once juicy profits of the early adopters.

The Solution

A library that knows how to keep a connection to a variety of exchanges and provides a platform to develop latency-sensitive trading algorithms is the problem that the Crypto Currency Exchange Connector is attempting to solve.

The library is being written in C++. It will be modular to support features such as persistence, risk management systems, and the FIX protocols. But the bare-bones system will get the information from the exchange and present it to the strategy as quickly as possible. It will also take orders generated by the strategy and pass it to the exchange as quickly as possible. A good amount of effort has been put forth to make those two things performant.

Giving strategies the information needed to make quick decisions was the focus. Keeping strategies from the need of “boiler plate” code was a secondary goal that I feel I accomplished.

Compiling in only the pieces necessary was another focus. I debated for a good bit about making each exchange a network service you can subscribe to. That makes sense for a number of reasons, but it is an extra latency that I did not want to pay for. And having everything compiled into one binary does not mean it is impossible to break it out later. I wrote it with the idea that off-loading exchanges to other processes or servers would be possible, but having it all-in-one is the default.

Persistence is another issue. There is a ton of data pouring out of the exchanges. Persisting it is possible. But having it in a format for research often requires dumps, massages, reformatting, and sucking it into your backtesting tool. To alleviate some of those steps, I chose TimescaleDB. It is based on Postgres, and seems to work very well. Dividing tick data into chunks of time can be done at export time. I have yet to connect it to R or anything like that, but I am excited to try. That will come later.

Project Status

I truly do not want this project to be set aside yet again. I have a strong desire to get this across the finish line. And it is very close to functional. There are only a few pieces left to write before my first strategy can go live.

I will not be making this project open source. It is too valuable to me, and gives me an edge over my competitors. At best I will take the knowledge I’ve gained and apply it to other systems that I am asked to build.

Arduino and the Arduino IDE are great products. They serve as an on-ramp (and a very good one) for many makers. You can do quite a bit of tinkering, and even build commercial products without leaving the Arduino ecosystem. I give the many developers, volunteers, makers, etc. big kudos for their work.

It is just not for me. I’ve never been one that likes having development steps done for me. I need a lower level. I need to see the gears (rusty or greasy as they may be) turn. So an IDE that does everything for you is great. But at times it gets in the way. And that is when I start to look under the hood.

Let me be clear. I will probably never design my own chip. I am also not one to dig too far in the minutiae of manufacturer’s data sheets. I like a certain level of abstraction. But when something is not working as I expect, my instinct is to check my setup. And when the setup is all done for me, I quickly start groping at straws, and often waste time.

A recent experience solidified my desire to leave the Arduino IDE behind and go it alone. And by doing so, while it was investment of time, added significantly to my repertoire of skills. And it is something I believe is a worth-while investment for many developers.

In my case, I was not working on an Arduino. I was working with a chip and associated hardware that had been twisted to work with the Arduino IDE. That was a great way to get started. Again, a big thank you to all that worked on making that work. But when things get “twisted” to make it work, often things get lost, dropped, or forgotten.

Fortunately in my case, the manufacturer of the micro controller has a well-documented SDK and gcc toolchain. And while I have yet to remove all Arduino pieces of my code, (and may never do so) I can say I am now much more confident in what will happen when I tweak my linker script. I know what goes where and when it will be called. When all of the sudden a reboot is triggered, I am usually certain as to what area of the code is in need of attention.

If you know how to recover from accidentally “bricking” your device, I believe you will love stepping outside of the Arduino IDE and into the world of tool chains, linker scripts, master boot records, and boot loaders. Yes, it is frustrating at first, but the feeling of success at the end often makes it worth it.

That is my $0.02. As always, feel free to chime in with your opinion.

There are plenty of avenues to enjoy some good old-fashioned coding. Lately, I’ve been doing some embedded work, and found a reason to explore smart watches. It seems quite a community is building around watches and fitness bands.

I purchased an inexpensive watch model that seems to have some grass-roots support around it. The P8. It was so inexpensive, I took a chance on its successor, the P9. They arrived today. The brain-dump and camera-dump will be built up here.

I have always enjoyed electronics as a spectator. It continues to fascinate me. But the maturing trend of IoT captures my imagination because it blends electronics and software.

My career has mainly revolved around the major platforms. Windows, MacOS, Linux. I wrote large applications that ran on large servers or powerful desktops. Performance was a concern, but the cost of development often demanded to worry about hardware constraints only when something didn’t work.

Smartphones and other mobile devices drove developers to once again consider the hardware. Sure, there are development tools that hide some of the complexities of particular hardware. But when you get down to a board that has few microprocessors, limited RAM, little storage, and must run on a watch battery, the idea of adding a tool to hide complexities often means the tool is too big to fit on the device, let alone your code.

Robots have been used as a STEM tool in schools for a while now. There are people in the workplace now that have a much more solid base in electronics than I do. Imagine kids, my school had a computer in the library that had to be shared among the entire student body. Of course, I was one of the few that knew how to turn it on. But I digress..

The idea that the latest generation (and even a few before this one) are being exposed to IoT makes me happy. Learning such things while your mind is open to explore (and you have the time to do it) means more capable people with imaginations beyond the basics.

But that doesn’t mean that people from my generation can’t contribute. Just as the generation before us, we can certainly understand that we are where we are because of the shoulders we stood on. The question I ask myself is “How can I help?”

I’m no writer, and I’m no teacher. But I do have a good amount of experience, much of which only exists as anecdotes in my head. So it is time I do more thinking, learning, and writing.

My hope is that by writing more, I will explore more. I am looking toward IoT to expand my electronics knowledge, increase my coding knowledge, and learn how to best explain concepts.

These are lofty goals, but I’m hoping for the best.

I am researching open-source projects for marine weather radar.

The weather radar hardware manufacturers are proprietary. I understand why, and I’m not against it. But I would like to have my data in one place if I can. To do that, I need integration. That will affect my choice in purchasing hardware.

To research:

Hardware manufacturers of marine radar for pleasure craft. Do they provide an API? What equipment is necessary?

Open-source projects used for integration of systems in the marine environment.

What I’ve found:

OpenCPN seems to be the open-source software of choice for chart plotter navigation. It seems they have AIS (traffic) integrated among others. I have also seen integration with older radomes as part of the radarpi project.

The industry standard for networking devices together is NMEA2000, which is a protocol far too slow for radar data, but seems to work well for things like wind speed, gps position, depth, etc.

Signal K is an open source project for integration of NMEA2000 stuff into the PC. From what I’ve seen so far, they’ve put a lot of work into it.

The Ultimate:

A PC connecting to the boat’s WiFi network that connects to a server (perhaps running on a Raspberry Pi) that talks to boat systems.

Related links:

GnuRadar: Need to explore

Cruiser’s Forum: Post about someone wanting to do the same, but dated.

SI-TEX MDS 8R was (is?) a Radome that plugged into a PC’s ethernet port. Interesting, not sure if it continues to be in production.

After more research, it seems radar is going the ethernet route. More of them are going headless, and I am hopeful that standardization is coming soon.

Update 16-September-2020

radar_pi has done a lot of work on this already. It seems the tight grip that manufacturers have on their hardware limits (but does not eliminate) the usefulness of writing such code.

Manufacturers do sell the antenna separately from their head units. Whether a setup replacing their head unit with a generic one would cause things like voided warranties have yet to be seen.

To truly reverse engineer, build, and test such software would require the actual hardware. This post talks about someone working on such a thing with Furuno hardware.