Thursday, September 07, 2017

John Sloan on Time and Space

Once again my friends and colleagues at Gogo Business Aviation in Broomfield Colorado indulged me, in the form of my alter ego John Sloan, by letting me come and give a talk entitled Time & Space on the historical connection between precision timekeeping and navigation. They taped it and put it on the 'tube. I had a great time, and hopefully I wasn't the only one. (I did, in my enthusiasm, misspeak a couple of times, and have made come corrective comments on the YouTube page.) Among lots of other things, I describe how both GPS and atomic clocks work.

A big thank you to the folks at Gogo for sponsoring this.

Wednesday, July 12, 2017


I have a tough time thinking of the Internet of Things as a new technology arena, considering I've worked on all kinds of embedded systems that were connected wirelessly, directly or indirectly, to the Internet. These ranged from eight-bit micro controllers to in-flight entertainment systems for business jets.

Recently I've been futzing around with a couple of IoT devices:


the Amazon Web Services IoT Button and


the TP-Link Smart Plug Mini. I was inspired by the clever work of my friend, former colleague, and one time college flat mate Paul Moorman who used the IoT Button with a Smart Plug to implement a wireless light switch. Based on prior work by other folks who reverse engineered the Smart Plug (and which he referenced in his article), Paul controlled the Smart Plug by punching a hole in his firewall for the port on the Smart Plug from which it receives commands, 9999, and then writing an AWS Lambda script in Python to talk to it. Pressing the IoT button stimulates the Lambda script in the AWS cloud to send the appropriate message to the Smart Plug to turn the power to any connected device on or off. Pretty cool.

I'd be lying if I didn't admit I was a little envious of Paul's ingenuity.

Ultimately I came up with my own practical application - significantly, as it turned out, one that required no development or firewall changes on my part - of just the Smart Plug: to turn our "cat cam" - an existing web cam in our kitchen near the cats' food bowls - on and off from my iPhone. It was pretty trivial to do. On a recent trip from Denver to Phoenix I was able to turn the cat cam on remotely using the TP-Link Kasa app and then use the webcam's app to peek at our beloved feline overlords (long may they rule).

Here's the thing: I had to punch a hole in my firewall for a port forwarding rule to connect to the cat cam from the webcam app on my phone, resolving the address using dynamic DNS. But I didn't have to punch a hole for the Smart Plug, or use dynamic or any other kind of DNS identifying my home network. What had to be happening is that the Kasa app was connecting to a server in the inter webs, and the Smart Plug was making its own outgoing bidirectional TCP connection through my firewall to a server (perhaps the same one) also on the inter webs.


My first stab at reverse engineering this was to dig out my AirPcap NX tool, a USB 2.0 device from Riverbed Technologies that allows you to look at WiFi traffic and, providing you have the WPA pass phrase and catch the wireless end point when it first connects to the WiFI network, decrypt and decode IP packets as they fly through the air. The AirPcap has drivers for Windows and it is recognized by Wireshark.

This pretty much went according to plan until I realized that I was only seeing the traffic from the server on the inter webs to the Smart Plug, not in the other direction. Peering closely at the TCP connection setup responses from the server, I could tell definitively that the Smart Plug was the connection instigator. The Smart Plug was establishing a TCP stream socket to port 50443 on server That domain name resolved to a server that was hosted in the AWS cloud (and unrelated to the AWS IoT button implementation). I could see packets from the server to the Smart Plug, but not in the other direction.

This put me in a paroxysm of web searching.

What I think was happening (see disclaimer above) is that my home WiFi access point supports 3 x 3 MIMO (multiple input multiple output) WiFi channels: it exploits spatial multiplexing to divide the packet stream into three transmit and three receive WiFi streams, all on the same radio channel, and which are all in the air simultaneously, each being handled by a separate antenna. My AirPcap NX supports at most 2 x 2 MIMO; it simply could not see all the traffic in the air.

I was using the AirPcap on my company laptop, a Lenovo ThinkPad running Windows 7. As part of my research, almost by accident, I came across an article describing the WiFi radios used by Apple, and discovered that my personal MacBook Pro came equipped with a radio capable of 3 x 3 MIMO with the same WiFi channels and technologies as my access point. Even better, the MacOS WiFi drivers natively support "monitor" mode, equivalent to promiscuous mode for Ethernet wired interfaces. And Wireshark for MacOS understood all of this already.

It didn't take very long for me to verify all of this using my MacBook Pro. I had Wireshark on my personal laptop peeking at all of the WiFi traffic between the Smart Plug and the TP-Link server in the Amazon cloud.

Spur WLAM pcapng

I was pretty proud of myself until Paul asked me why I didn't just insert a SharkTap into the Ethernet connection between my cable modem and my access point and Wireshark the traffic from there. A little embaressed, I did that too.

SharkTap Between Cable Modem and Router

For sure that was the simpler solution, but it had the side effect of NATting all the IP addresses on my home network to the one address assigned via DHCP to my access point by my internet service provider. It made the Wireshark traces a little harder to grok, but really no big deal.

Never the less, I'm glad I wasn't as smart of Paul and so didn't think of that solution right off the bat. I now know I have another useful tool in my troubleshooting arsenal, one that came ready made right out of the box.

Tuesday, June 13, 2017

Practical Matters

I'm frequently working on prototype hardware whose design bears only a superficial resemblance to what we are eventually going to ship. Sometimes there is a couple generations of prototypes in very different form factors, with the final version arriving not much before our ship date. Prototypes give the software team a head start (the header, the better) while the hardware team (with a little help from guys like me) work out the bugs in the hardware design. I may find myself rolling an engineering model running on batteries around in a parking lot to test that the cellular modem transitions seamlessly between cell towers. Or crawling around in a twin-engine executive business jet trying to figure out why the satellite communications unit isn't coming up. Or sitting on the floor in a drafty and noisy and sometimes dark server or telecom room between equipment racks, tangled in a maze of cables, with a laptop in my lap.

I love doing this kind of work. But it does require a fair amount of flexibility and no small amount of improvisation on my part. Here - in no particular order - are some of the tools I've found useful over the past few years. Maybe you will too.

Cheater Safety Glasses
Disposable Ear Plugs


The LASIK surgery I got eons ago was a miracle; I still have 20/20 distance vision. But the older I get, the worse my close up vision becomes. So besides protecting my eyes, these safety glasses, that have reading glasses molded into them at the bottom of each lens, let me see what the display on that SATCOM box in the equipment bay is actually telling me. I keep a pair of these in just about every tool kit I have, in my daily briefcase, and in the garage at home. I buy 150s (1.5 diopter) currently, but alas soon to be 175s, at my local hardware store.

My hearing, which has never been very good (a youth spent target shooting, riding motorcycles, and working in white-noisy computer and equipment rooms), hasn't gotten any better. I try to protect what little I have with disposable foam ear plugs. I have several different kinds laying around, but these E.A.R. brand Taperfit with the lanyard work best for me for day to day work. I use them in the lab, in the equipment and server rooms, and on site like when the pilot fires up the auxiliary power unit. Sometimes I use them at the coffee shop. You can buy these at your local hardware store too, but I order the big industrial sized box of individually wrapped ones from Amazon and hand them out (some might say forcibly) to co-workers.

Roll Your Own Custom Outlet Strip


I buy these little outlet expanders and grounded switches and when circumstances require I fabricate my own outlet strip on the spot. The outlet expanders play well with power bricks, and have a little light to indicate that they are electrified. You can place the switches wherever you need to control power to an individual device or set of devices. Add a heavy duty extension cord, and you have a really useful and flexible piece of electrical power distribution kit.

Folding Stool


My knees are getting too old to squat on the floor as I peer at some piece of gear. So I buy these folding stools. Sometimes I sit on them, sometimes I sit on the floor and use them as a laptop stand. They are sturdy, can handle my manly weight, are easy to schlep around, and when folded don't take up much room. I order them from Amazon, but I've seen something similar at my local hardware store.

Aluminum Project Clipboard



When I head to the hardware store to buy something to solve some problem in the lab, I don't want to find out I should have brought a tape measure. Or reading glasses. Or writing material. Or a compass (seriously). I find these aluminum project clipboards, beloved by state highway patrol officers everywhere, really useful. They are pretty indestructible, and hold a bunch of useful stuff inside. Also useful in the field, for all the same reasons. I got this one at my local hardware store.

Velcro Cable Ties


I don't know how I lived without these. When I get a new cable - ethernet, USB, video,  electrical, doesn't matter - the first thing I do now is toss the useless wire cable tie and replace it with one of these velcro cable ties. They come on continuous perforated spools of 25, so you just unwrap one and tear it off. They have a loop hole on one end so you can lasso them to one end of the cable, or connect several together when you need a longer cable tie. They can be easily trimmed with scissors to make them shorter and easier to deal with. My only complaint is they only come in black or gray; I'd like something like safety orange or high-visibility yellow or even shocking pink. I buy them at my local office supply store.

Zip-lock Bags


Embedded development means having  a lot of small related parts and other crap you have to keep track of. I like these Hefty-brand bags with the sliding zip-lock for that. They come in lots of sizes; I have at least one box of each. They have a label area that takes well to writing on with a Sharpie. They are cheap and sturdy. I keep everything from connectors to a Raspberry Pi with its power supply in them. I buy these at my local grocery.

Quirky Cordies Cable Organizer


I have a dozen of these Quirky Cordies cable organizers laying around or permanently assigned to specific tasks. You place cables in between the upright loopy fingers. The Cordies are weighted in the base so they tend not to get pulled off lab benches or window sills (antennas, you know) by the weight of the cables they are holding. They come in a variety of colors (although again not safety orange or high-visibility yellow). I order these off Amazon. I can see seven of them just as I sit here writing this at my lab bench in my home office.

Improvised Ground
Anti-Static Strap


Ideally when you use an anti-static strap - you are using an anti-static strap, aren't you? - you want to ground yourself to the chassis or grounding point of whatever you are working on. Easy in the lab, sometimes not to simple in the field. Although it's not the perfect solution, I came up with these improvised grounds by buying three-prong plugs, and grounding wires (package of two), at my local hardware store. Five minutes spent putting them together makes for an improvised ground in any grounded outlet I plug it into. You do have to be aware of the possibility of ground loops if the stuff you are working on isn't plugged into the same electrical circuit, so some caution is called for.

Although any anti-static strap is better than no anti-static strap, I like these with the Twist-O-Flex bands. I buy them off Amazon. The alligator clip pulls off to reveal a male connector compatible with the female plugs on the anti-static mats I like to use.

Thermal Label Maker


Folks that work with me will tell you I am an obsessive-compulsive labeler. That's why when the quality process inspector comes by to do an audit, it's my lab bench they take them to. I also add my email address and phone number to most of the equipment my company owns in addition to an official property tag. Lots of folks make these thermal label makers, all of which use proprietary cassettes containing a spool of thermal label tape, and which typically come in a variety of foreground and background colors. I think Brother may have discontinued this particular model, which I like because it is capable of printing two lines of text. But once you get bit by the labelling bug, you may start labelling stuff around the house. I bought this one from my local office supply store.

Carlisle Folding Cart



The company that makes this folding cart, Carlisle, sells a variety of sizes and shapes, mostly I suspect for the food service and catering industry. Turns out it's also really useful for field testing. I like this medium-sized but heavy-duty one because it's big enough to carry some serious systems under test, but folds up small enough to fit in the back of my car. It has locking wheels, so you don't see your twenty thousand dollar prototype roll away across the parking lot.

USB LTE Cellular Modem


Admittedly, this is an expensive piece of kit. But when I've needed it, I've really needed it. I've used this Verizon MiFi U620L LTE modem to get an internet connection on a Windows laptop when working in an airplane hanger with no WiFi. Or on a Mac laptop while parked in my car. Or on a Raspberry Pi running on a battery pack and sending telemetry to the mothership. I got this one from Verizon and added it to my cellular phone plan. The incremental cost to my plan was relatively tiny, although the device itself was pretty pricey. But sometimes it's actually faster than the local WiFi, and likely more secure.

Raspberry Pi
Raspberry Pi TTL Serial to USB Adapter

Raspberry Pi 3 "tin"


The Raspberry Pi single board computer (SBC) is frequently useful for early prototyping of the more powerful ARM-based embedded systems. I keep one attached to my network in my home office just so I can do quick unit tests on a 64-bit ARM target.

The FTDI TTL-232R-RPI debug cable is an off the shelf serial console to USB adapter that uses my favorite USB-to-serial solution, the FTDI chip. You can order these off Amazon. I added the "BYO 68A" label, which reminds me that the black wire goes to pin 6, the yellow wire to pin 8, and the orange wire to pin 10, on the Pi.

Ginormous Lithium Uninterruptible Power Supply



Another expensive piece of kit, this InteliCharge X1300 UPS has a big lithium battery pack that provides up to 1150Wh to 110V or 220V mains outlets, plus two USB outlets, and a 12V cigarette lighter outlet, for your portable power needs. I've run a large embedded system intended for aircraft power for hours on this unit. I bought mine from Amazon, although I notice they no longer list it; a few moments of web searching suggests it's no longer available. I have a smaller 400Wh unit from Anker that I have also used and which would be my second choice. I've used the InteliCharge to test equipment in a Faraday cage, to roll around in a parking lot to test cellular connectivity, or to move a piece of equipment around in a building without having to power it down.


This photograph shows the UPS and at least four other items from this article in use at a client's site. The UPS is powering the system under test, a laptop, and a VOIP phone. That is indeed Gigi the cat from the anime Kiki's Delivery Service on that bag on the bottom shelf in the photograph above.

USB Charger/Battery

Anker PowerCore Fusion 5000 Charger/Battery

On a much smaller scale, devices like this Anker PowerCore Fusion 5000, which combines a wall wart USB charger with a 5000mAh USB battery, are just the thing for powering the Raspberry Pi and similar USB-powered devices off the grid.

Hazer sender "tin"

Here is the PowerCore plus at least five other items from this article. The battery will power the Raspberry Pi so that it can send geolocation telemetry via the LTE modem back to the mothership.

Important safety tip: The PowerCore Fusion 5000 is not a uninterruptible USB power supply. Disconnect the PowerCore from the mains electrical outlet and anything connected to it will notice a power disruption while the device switches to its internal battery.

NRT ViperBoard


The embedded systems I help my clients develop always have scads of general purpose input/output (GPIO) pins used to communicate with or control sensors or devices. But often, most of the application software the team develops could be tested on a desktop personal computer or even a laptop long before a laboratory prototype is available. But such mainstream systems typically have no GPIO pins. The ViperBoard from Nano River Technologies (NRT) solves that problem. Attach a ViperBoard via USB to your PC and suddenly you have a bunch of GPIO pins, as well as I2C and SPI serial busses. This is also useful for interfacing your PC to lab test equipment. The ViperBoard is supported right out of the box by the more recent Linux kernels. (I added the plastic standoff legs and the "2" label.)

GlobalSat USB GPS Receiver


Similarly, for products which will eventually feature geolocation, a lot of the application software could be tested on a desktop or laptop if only GPS were available. There are lots of USB GPS receivers (and I've tried most of them), but my go-to unit is the GlobalSat BU-353S4. Plug it into a USB port and it enumerates as a serial port. Read from the serial port and you get a stream of standard NMEA sentences providing latitude, longitude, time, and date.

USB to Serial to USB Adapter


I've written before about my deep love for the CableMax USB 2.0 to RS-232 DB-9 Adapter, which contains an FTDI chip, my favorite USB to serial converter. Here's an application of it that has turned out to be remarkably useful: I attached two adapters back to back with a null modem converter in between, making a device that can connect two computers via a serial connection, all via USB. This may sound mundane, but lots of embedded application software still make use of serial ports to talk to devices. This little piece of kit lets you test such software on your desktop or laptop without a serial port in sight.

SharkTap (updated 2017-06-28) (updated 2017-07-11)


I've mentioned this device before, but it bears repeating. The SharkTap is an Ethernet tap designed for running Wireshark traces on a LAN. It supports ten megabit, one hundred megabit, and gigabit Ethernet speeds. It passes through Power over Ethernet (PoE) so it can be used on the growing number of devices (like, for me, VoIP phones) that get their power over their Ethernet connection. It is itself powered via a USB cable.

SharkTap 10/100/1G Wireshark Network Tap

The SharkTap is conceptually dirt simple: it contains a three-port (at least) Broadcom Ethernet switch configured to pass through traffic in either direction on the (as shown here) blue and green Ethernet cables, while mirroring everything in both direction onto the red Ethernet cable.

SharkTap Between Cable Modem and Router

Here's a photograph I took just moments ago of me using the SharkTap to peek at the traffic between an IoT device - a TP-Link Mini Smart Plug - with its server - in the Amazon Web Services (AWS) cloud as it turns out - by tapping into the WAN traffic between a cable modem and a router/access point.


The SharkTap folks have introduced another model, the SharkTapUSB, that works identically to the one I describe above, but in addition has a built-in USB 3.0 to Gigabit Ethernet interface. This greatly simplifies using laptops which lack a wired Ethernet interface (i.e. pretty much all of them), while using the same USB 3.0 connection to power the SharkTap. It worked great with my MacBook Pro, eliminating the need for me to use my Apple Thunderbolt Gigabit Ethernet adaptor.

USB Power Supply (updated 2017-07-11)


If you're like me, you have a pile of USB power supplies laying around and you have discovered that not all USB power supplies are created equal. I like this one in particular, the Anker PowerPort 2. It provides a maximum of 2.4A to each of two USB ports (4.8A total). It's small and its electrical plug folds away, making it easier to cram into your field kit.

Lab Assistant


It's useful to have a lab assistant to help with all the chores that come with product development and to bounce ideas off of. I've had a lot of good ones over the years. Just make sure you teach them good electro-static discharge (ESD) practices. Here you can see one of my assistants using an anti-static mat.

Tuesday, May 30, 2017

Stranger Android Things

As far back as 2011 I was talking up using Android as a Linux-based embedded operating system. In those days I was running it on a BeagleBoard. I liked that idea then and I still like it now.

BeagleBoard xM Rev C

Android has a rich set of features, many of which would be really useful in an embedded product at the higher end. It is architected to make use of multiple network devices - for example, WiFi and a cellular modem - which is becoming common in the kinds of things I work on. Android has a mature tool chain and development infrastructure that includes stuff like a graphical IDE and a debugger. It is a well understood platform on which to deploy third-party applications, which at least one of my clients is eyeing as a business model for one of their products. And best of all, Android has enormous market forces incentivizing other folks to do a lot of the work for you.

I never got any traction on that idea. But maybe now that Google is promoting Android Things - as in Android for the Internet of Things - that will change. Android Things (formerly the platform known as "brillo") is a stripped down version of Android that can be run headless. It has the usual Android Run Time (ART) Java Virtual Machine (JVM) plus support for native C and C++ code, so you can port your legacy drivers and libraries, but do your new application development in an object oriented language that has first-class built-in support for threading, synchronization, and lots of other useful stuff.

Or will. Currently Android Things is just a pre-built developer preview that you can load on a selection of inexpensive development boards like the Raspberry Pi 3. It includes drivers and Java APIs for common embedded hardware interfaces like an Inter-Integrated Circuit (I2C) bus, a Serial Peripheral Interface (SPI) bus, Pulse-Width Modulation (PWM) output, and General Purpose Input/Output (GPIO) pins.

Third-party developers have in turn written higher level drivers and Java APIs on top of these for a selection of peripherals on commercially available expansion boards. For the Pi, that would be the Pimoroni Rainbow Hat. The Rainbow Hat has an APA102 LED strip (SPI), a BMX280 temperature and pressure sensor (I2C), two two-character HT16K33 alphanumeric segments (also I2C), a piezoelectric buzzer (PWM), and some discrete LEDs and buttons (GPIO). Below is a photograph of a Rainbow Hat on a Pi running my single threaded RainbowHatDemo application.


The source for Android Things hasn't been released yet, so porting any of my C or C++ libraries will have to wait. But the developer preview, with a Raspberry Pi and the Rainbow Hat, is a great way to see where this is all going. I wrote some toy applications using Android Studio just to try to remember how to use Java, which I haven't done anything serious with in a decade. You can find the repository, code-named Deringer, with all my source code as Android Studio projects, on GitHub. Below is a little code snippet from my multi-threaded RainbowHatThing application that reads the temperature and pressure sensor and logs the results.

    public void run() {
        Log.i(getClass().getSimpleName(), "begin");
        while (!lifecycleDone()) {
            try {
                float[] readings = sensor.readTemperatureAndPressure();
                float centigrade = readings[0];
                float fahrenheit = (centigrade * 9.0f / 5.0f) + 32.0f;
                float hectopascals = readings[1];
                float inches = hectopascals * 0.02953f;
                Log.i(getClass().getSimpleName(), centigrade + "C " + fahrenheit + "F " + hectopascals + "hPa " + inches + "in");
            } catch (IOException e) {
                // Do nothing.
        Log.i(getClass().getSimpleName(), "end");

Here is a little video (with audio) of the RainbowHatThing application running many threads concurrently on my tiny test system.

Will Android Things take off as a platform for the more capable IoT targets? No clue. But for sure there are many economic reasons to take this very seriously.

Monday, May 22, 2017

The Cost of Doing Business

My software development career over the past forty-plus years has been all over the map: supercomputers, eight-bit micro controllers, vast distributed telecommunications systems, multi-core microprocessors. The one thing they all have in common - at least at the level I work at near bare metal - is that the skill sets are all basically the same. But one thing that isn't the same is the cost of the tools to debug, functionally test, and validate my work.

Perhaps twenty years ago, I was writing firmware in C and C++ for optical networking products based on ATM and SONET. One night I found myself working late in a lab trying to debug some recalcitrant real-time code. I was surrounded by an H-P broadband analyzer, two H-P ATM protocol analyzers, and a tangle of multimode optical fiber. I realized I had maybe US$300,000 worth of test equipment dedicated to this one task, never mind my high-end workstation and the room-filling racks of telecommunications equipment. I paused for a moment to realize that an organization smaller than Bell Labs might have had a tough time tackling this product development effort, the investment in infrastructure was so significant.


Moore's Law hasn't really changed things that much, because as the tools got cheaper, the systems we develop and debug have gotten larger and more powerful right along with them. Just the other day I was working on my stratum-0 atomic clock project and I realized that my little hobbyist oscilloscope - one of the most useful tools I have ever purchased - was completely inadequate for doing any kind of jitter analysis on the output of the device. After some reflection, I decided I probably didn't know anyone that had an oscilloscope that could measure the jitter of my chip-scale atomic clock. Any jitter that might be detected would surely be in the oscillator used by the measurement instrument itself.


See, our measurement of time is a funny thing. It is a completely synthetic artifact, a singularly human invention. Thanks to natural variation in the Earth's rotation and in its orbit, time isn't even tied to the movement of the Sun any longer. It's defined, in the International System of Units, to the beat of an oscillator based on the cesium-133 time standard. And the thing I was trying to measure was itself an oscillator based on the cesium-133 time standard. The only way we can measure the precision and accuracy of a clock is by using a more precise and more accurate clock. I'm not likely to ever have one of those.

The larger lesson here is this: before you decide to tackle a project, make sure you can get the tools you need for the entire life-cycle of the product. You may find out that the economics of developing a product are substantially different from the economics of debugging, testing, validating, and supporting that product.

Wednesday, May 17, 2017

Engines of Time

In My Stratum-1 Desk Clock and My Stratum-0 Atomic Clock I described the two clocks that I built: O-1 a.k.a. "Hourglass", and O-2 a.k.a. "Astrolabe". Both act as stratum-1 NTP servers on my home network, along with two commercial units, "Waterclock" and "Sundial", that I have previously described. I mounted both clocks on inexpensive acrylic fixtures from office supply stores that just happened to be about the right size and shape to serve as stands to display my handiwork.

Hourglass on a Stand

Hourglass is my desk clock sitting on a shelf in my home office. It is a Raspberry Pi 3 using Raspbian and runs the GPS daemon and the NTP daemon. It connects to my home network wirelessly, although the wirefull port works as well. It uses the NMEA output and 1PPS strobe from a Uputronics GPS board to discipline its system clock. It has a real-time clock board that with the push of a button can save the time to a battery-backed clock to preserve its state should it need to be powered off. Thanks to the standard Linux kernel and GNU software, it automatically adjusts to Daylight Saving Time, and GPS itself automatically adjusts to the occasional leap second. As long as it can receive GPS signals from an amplified antenna sitting in my office window, it will be within milliseconds (at least) of Universal Coordinated Time.


Astrolabe sits on a small table in the living room. It is virtually identical to Hourglass in both hardware and software, but replaces the Uputronics GPS board with a Jackson Labs Technologies CSAC GPSDO, a board that has a Microsemi Chip-Scale Atomic Clock that it phase locks to GPS time. Astrolabe receives NMEA output from the GPS chip on the GPSDO board, but gets its 1PPS strobe from the CSAC cesium-133 oscillator. Once the CSAC is initially disciplined to the GPS 1PPS, and the Pi's system clock is disciplined to both GPS and 1PPS, even if it subsequently cannot receive GPS signals from the matchbook-sized amplified antenna sitting in the living room window, it will still be within milliseconds or even microseconds of UTC. Astrolabe also includes a couple of test points: a BNC connector to the 10 MHz output of the CSAC, and an RS-232 serial port to the command interface of the GPSDO.

This is the most accurate and precise timepiece I will ever own, perhaps could ever own, my own stratum-0 atomic clock. Far far better clocks exist, but I am unlikely to ever own them, much less have them in my living room. Its time is traceable to the atomic clocks in the U.S. NAVSTAR GPS satellite constellation in orbit, which are traceable to the U.S. Air Force Master Control Station clock in Colorado Springs Colorado, which is traceable to the Washington Master Clock at the U.S. Naval Observatory in Washington D.C., which is traceable to the NIST-F2 clock at NIST in Boulder Colorado, which is synchronized with atomic clocks all over the world coordinated by the International Bureau of Weights and Measures in Paris.

I have been fascinated by time, time measurement, and time keeping, for decades. Having spent the past forty-plus years working on real-time systems close to bare metal, this was perhaps inevitable. The ability for information to cross the boundary between the real world and the digital world fundamentally depends on the quality of our frequency sources. And our interpretation of reality rests on the accuracy and precision of our clocks.

Tuesday, May 09, 2017

My Stratum-0 Atomic Clock

In My Stratum-1 Desk Clock I wrote about my quest for accurate and precise timekeeping that lead me to hack together a desk clock that disciplined its system clock against time taken from the U.S. global positioning system (GPS). I have come to think of this first effort as "O-1", inspired by John Harrison, the eighteenth-century carpenter who invented the marine chronometer and many of the fundamental innovations that remain in quality mechanical clocks today. As detailed in Dava Sobel's excellent book Longitude: The True Story of a Lone Genius Who Solved the Greatest Scientific Problem of His Time, Harrison designed and built a series of chronometers which he named H-1 (for Harrison-1) through H-5, in competition for the prize offered by England's Board of Longitude, an organization that in many ways was the DARPA and NSF of its day. Some other horologists of his time adopted his nomenclature for their own entries into that competition.

But O-1 has a flaw: it is only as good as its signal from the GPS satellites. Should the GPS transmissions, from which the one Hertz 1PPS pulse was synthesized, be unavailable, due to weather or what not, O-1 would be only as good as the standard quartz oscillator used by its Raspberry Pi 3. In time, O-1 would drift away from UTC. What good is that? What I needed was a better, more stable oscillator.

What I needed for O-2 was an atomic clock.

Atomic clocks are not clocks in the sense of telling the time of day or measuring time duration. What they are are extremely stable frequency sources, a mechanism to generate very regular ticks, tied to a fundamental physical property of the universe. Modern atomic clocks have the usual trade off of cost versus accuracy and precision, so that they can range from the big refrigerator-sized units fielded by the U.S. National Institute of Standards and Technology (NIST) costing hundreds of thousands of dollars, to the suitcase-sized commercial units found in telecommunications systems and which cost tens of thousands of dollars (and which from time to time you can find used on eBay for around U.S. twelve to fifteen grand).

All of those were outside my price range.

In 2001, and again in 2008, emboldened by developments in practical manufacturing of micro-electro-mechanical systems (MEMS), and by research by organizations like NIST into exploiting quantum effects to eliminate some of the bulkier portions of classic atomic clock designs, the U.S. Defense Advanced Projects Agency (DARPA) notified the R&D community, in BAA-01-32 and BAA-08-32 respectively, that it was interested in funding research in, and development of, a chip-scale atomic clock (CSAC). A CSAC would be an atomic clock - specifically, a cesium-133 reference oscillator - packaged in a surface-mount device.

The implications of such a device are wide ranging and profound. Precision frequency sources are ubiquitous components in both military and civilian applications in the realm of telecommunications, sensors and instrumentation, navigation, and even munitions. A relatively inexpensive, easily portable, and - it must be said - economically destructible cesium reference oscillator would open up vast possibilities of new mobile devices requiring precision timing and frequency standards.

Your tax dollars at work: San Jose-based Symmetricom, who had already purchased the frequency and time standard business unit of Agilent, formerly Hewlett-Packard, the folks who made the suitcase-sized cesium atomic clocks, developed the SA.45s CSAC. (Symmetricom has since been acquired by its California neighbor Microsemi.) Symmetricom with Las Vegas-based Jackson Labs Technologies, a developer of GPS-disciplined oscillators (GPSDO), integrated the CSAC "physics package" (as it is euphemistically called) with a board that incorporates a u-blox GPS chip and a 32-bit ARM-based microcontroller, to create a CSAC GPSDO. This provides the time of day via the output of the GPS chip, and phase locks the cesium oscillator with the 1PPS derived from the GPS system.


Once the output of the CSAC is initially disciplined against GPS, the board can maintain better than 0.3 parts per billion stability in the absence of a GPS signal. An O-2 built with the JLT CSAC GPSDO could be within a few microseconds of UTC per day, even without GPS.

It sure wasn't cheap, but it was doable.


O-2 runs virtually the same software as O-1, with only minor tweaks. The function of the O-1 GPS board has been replaced by the JLT GPSDO, but the interface to the Raspberry Pi 3 is the same: NMEA 0183 sentences arrive via the serial port, and the 1PPS via a GPIO pin. The interface between the two boards was only complicated by the need for level translation: from RS-232 on the GPSDO to TTL on the O-2 for NMEA, and from 5V logic on the GPSDO to 3.3V logic on the O-2 for 1PPS.

The ARM microcontroller on the GPSDO has its own LCD display (on the right) in addition to the O-2 LCD display (on the left). Here I show it - somewhat redundantly - displaying the time, but a button on the GPSDO can be used to cycle through a number of status displays.

The GPSDO has two serial ports, one read-only providing the raw NMEA stream from the u-blox chip which I use to drive the GPS daemon on the Raspberry Pi, and one read-write serial port used for ASCII commands to the ARM microcontroller on the GPSDO to which I attached a FTDI serial-to-USB convertor connected to my desktop. The GPSDO exports a second serial control interface via its own FTDI chip over the USB connection to the Raspberry Pi that is also used to power the device.

Like O-1 (a.k.a. "hourglass"), O-2 (a.k.a. "astrolabe") doubles as a stratum-1 NTP server on my local area network, accessible via either WiFi or by a CAT5 cable.


(I now have four stratum-1 NTP servers on my LAN, all requiring GPS antennas. I dunno, is that too many?)

Four GPS Antennas

O-1 got mounted on a fixture to become a desk clock in my home office, costing hundreds of dollars in materials. I anticipate O-2 similarly becoming a mantle or wall clock, costing thousands of dollars. But both projects were sure awesome learning experiences.


Thanks to reader Fazal Majid who shamed me into developing the O-2. Thanks also to Said Jackson and the folks at Jackson Labs Technologies who graciously put up with some questions.




R. L. Beard, J. D. White, J. A. Murray, "Military Applications of Time and Frequency", Proc. of the 28th Annual Precise Time and Time Interval Systems and Applications Meeting, Reston, Virginia, 1996

Jackson Labs, CSAC GPSDO User Manual, ver. 1.6, 80200506, 2014-03-16

Jackson Labs, CSAC GPSDO Release Notes, rel. 1.2, 80200505, 2012-01-30

Jackson Labs, CSAC GPSDO Operating Recommendations, rev. 1.1, 2016-04-13

J. Jespersen, J. Fitz-Randolph, From Sundials to Atomic Clocks: Understanding Time and Frequency, 2nd ed., Dover, 1999

A. Lal, "Integrated Micro Primary Atomic Clock Technology (IMPACT)", DARPA BAA-08-32, 2008-04-21

Microsemi, SA.45s Chip-Scale Atomic Clock User Guide, rev. C, 098-00055-000, 2016-07

G. Miller, E. Raymond, GPSD Time Service HOWTO, ver. 2.10, 2016-09

E. Raymond et. al, Building GPSD from source, 2016-04-10

E. Raymond et. al, Hacker's Guide to GPSD, 2016-09-02

K. Shenoi, "Clock, Oscillators, and PLLs: An introduction to synchronization and timing in telecommunications", Workshop on Synchronization in Telecommunication Systems, San Jose, CA, 2013-04

D. Sobel, Longitude: The True Story of a Lone Genius Who Solved the Greatest Scientific Problem of His Time, Bloomsbury USA, 1995

Symmetricom, "Introduction to Symmetricom's QUANTUM(tm) Chip Scale Atomic Clock SA.45s CSAC", 2012 International Technical Meeting of The Institute of Navigation, Newport Beach, CA, 2012-01

Symmetricom, The SA.45S Chip-Scale Atomic Clock, 2011 Sanford PNT Symposium, Menlo Park, CA, 2011-11-18

W. Tang, "A Chip-Scale Atomic Clock", DARPA BAA-01-32, 2001-07-06

Tuesday, April 04, 2017

Some Stuff That Has Worked For Me In C

Diminuto (Spanish for "tiny") is a library of C functions and macros that support the kind of low level embedded, real-time, and systems programming work I am typically called upon to do on the Linux/GNU platform. It's no longer so tiny. I started it way back in 2008 and it is still going strong. Its open-source (FSF LGPL 2.1) code has found its way into a number of shipping commercial products for more than one of my clients. Diminuto can be found on GitHub along with eighteen (so far) other repositories (some private) of my work.

My forty year career has been a long, strange, and marvelous trip. Along the way, I've found a number of techniques of architecture, design, and implementation in C and C++ that have worked well for me, solving a number of recurring problems. I don't claim that these techniques are the best, or that I am the only, or even the first, developer to have thought them. Some of them I shamelessly borrowed, sometimes from other languages or operating systems, because I try to know a good idea when I see one.

This article documents some of those ideas.

Standard Headers and Types

Just about every big closed-source project I've worked on defined its own integer types. Even the Linux kernel does this. Types like u32 for unsigned 32-bit integers, or s8 for signed 8-bit integers. Historically - and I'm talking ancient history here - that was a good idea, way back when C only had integer types like int and char, and there weren't dozens and dozens of different processors with register widths that might be 8, 16, 32, or 64 bits.

But this kind of stuff is now available in standard ANSI C headers, and Diminuto leverages the heck out of them.

stdint.h defines types like uint32_t, and int8_t, and useful stuff like uintptr_t, which is guaranteed to be able to hold a pointer value, which on some platforms is 32-bits, and on others 64-bits.

stddef.h defines types like size_t and ssize_t that are used by a variety of POSIX and Linux systems calls and functions, and are guaranteed to be able to hold the value returned by the sizeof operator.

stdbool.h defines the bool type and constant values for true and false (although, more about that in a moment).

In at least one client project, I talked them into redefining their own proprietary types via typedef to use these ANSI types, which simplified porting their code to new platforms.

Boolean Values

Even though I may use the bool type defined in stdbool.h, I don't actually like the constants true and false. For sure, false is always 0. But what value is true? Is it 1? Is it all ones, e.g. 0xff? Is 9 true? How about -1?

In C, true is anything that is not 0. So that's how I code a true value: !0. I don't care how the C standard or the C compiler encodes true, because I know for sure that !0 represents it.

I normalize any value I want to use as a boolean using a double negation. For example, if I have two variables alpha and beta that are booleans. Do I use
(alpha == beta)
to check if they are equal? What if alpha is 2 and beta is 3? Both of those are true, but the comparison will fail. Do I use
(alpha && beta)
instead? No, because I want to know if they are the same, both true, or both false, not if they are both true. If C had a logical exclusive OR operator, I'd use that - or, actually, the negation of that - but it doesn't. I could do something like
(((alpha && beta) || ((!alpha) && (!beta)))
but that hurts my eyes.

Double negation addresses this: !!-1 equals !0 equals whatever the compiler uses for true. I use
((!!alpha) == (!!beta))
unless I am positive that both alpha and beta have already been normalized.

I also do this if I am assigning a value to a boolean
alpha = !!beta
unless I am very sure that beta has already been normalized.

Inferring Type

If I want to check if a variable being used as a boolean - no matter how it was originally declared - is true, I code the if statement this way.
if (alpha)
But if the variable is a signed integer and I want to know if it is not equal to zero, I code it this way.
if (alpha != 0)
And if it's an unsigned integer, I code it this way.
if (alpha > 0)
When I see a variable being used, I can usually infer its intended type, no matter how it might be declared elsewhere.

Parentheses and Operator Precedence

You may have already noticed that I use a lot of parentheses, even where they are not strictly speaking necessary. Can I remember the rules of operator precedence in C? Maybe. Okay, probably not. But here's the thing: I work on big development projects, hundreds of thousands or even millions of lines of code, with a dozen or more other developers. And if I do my job right, the stuff I work on is going to have a lifespan long after I leave the project and move on to something else. Just because I can remember the operator precedence of C, the next developer that comes along may not. So I want to make my assumptions explicit and unambiguous when I write expressions.

Also: sure, I may be writing C right now. But this afternoon, I may be elbow deep in C++ code. And tomorrow morning I may be writing some Python or Java. Tomorrow afternoon, sadly, I might be hacking some JavaScript that the UI folks wrote, even though I couldn't write a usable line of JavaScript if my life depended on it. Even if I knew the rules of operator precedence for C, that's not good enough; I need to know the rules for every other languages I may find myself working in.

But I don't. So I use a lot of parentheses.

This is the same reason I explicitly code the access permissions - private, protected, public - when I define a C++ class. Because the default rules in C++ are different for class versus struct (yes, you can define access permissions in a struct in C++), and different yet again for Java classes.

Exploiting the sizeof Operator

I never hard-code a value when it can be derived at compile time. This is especially true of the size of structures or variables, or values which can be derived from those sizes.

For example, let's suppose I need two arrays that must have the same number of elements, but not necessarily the same number of bytes.
int32_t alpha[4];
int8_t beta[sizeof(alpha)/sizeof(alpha[0])];
The number of array positions of beta is guaranteed to be the same as the number of array positions in alpha, even though they are different types, and hence different sizes. The expression
divides the total number of bytes in the entire array with the number of bytes in a single array position.

This has been so useful that I defined a countof macro in a header file to return the number of elements in an array, providing it can be determined at compile time. I've similarly defined macros like widthof, offsetof, memberof, and containerof.

Doxygen Comments

Doxygen is a documentation generator for C and C++, inspired by Java's documentation generator Javadoc. You write comments in a specific format, run the doxygen program across your source code base, and it produces an API document in HTML. You can use the HTML pages directly, or convert them using other tools into a PDF file or other formats.

 * Wait until one or more registered file descriptors are ready for reading,
 * writing, or accepting, a timeout occurs, or a signal interrupt occurs. A
 * timeout of zero returns immediately, which is useful for polling. A timeout
 * that is negative causes the multiplexer to block indefinitely until either
 * a file descriptor is ready or one of the registered signals is caught. This
 * API call uses the signal mask in the mux structure that contains registered
 * signals.
 * @param muxp points to an initialized multiplexer structure.
 * @param timeout is a timeout period in ticks, 0 for polling, <0 for blocking.
 * @return the number of ready file descriptors, 0 for a timeout, <0 for error.
static inline int diminuto_mux_wait(diminuto_mux_t * muxp, diminuto_sticks_t timeout)
    return diminuto_mux_wait_generic(muxp, timeout, &(muxp->mask));

Even if I don't use the doxygen program to generate documentation, the format enforces a useful discipline and consistent convention in documenting my code. If I find myself saying "Wow, this API is complicated to explain!", I know I need to revisit the design. I document the public functions in the .h header file, which is the de facto API definition. If there are private functions in the .c source file, I put Doxygen comments for those functions there.

Namespaces (updated 2017-04-25)

Most of the big product development projects I've worked on have involved writing about 10% new code, and 90% integrating existing closed-source code from prior client projects. When that integration involved using C or C++ code from many (many) different projects, name collisions were frequently a problem, either with symbols in the source code itself (both libraries have a global function named log), or in the header file names (e.g. both libraries have header files named logging.h).

In my C++ code, such as you'll find in my Grandote project on GitHub (a fork from Desperadito, itself a fork from Desperado, both since deprecated), there is an easy fix for the symbol collisions: C++ provides a mechanism to segregate compile-time symbols into namespaces.

namespace com {
namespace diag {
namespace grandote {

 ::pthread_cond_init(&condition, 0);

Condition::~Condition() {

int Condition::wait(Mutex & mutex) {
 return ::pthread_cond_wait(&condition, &mutex.mutex); // CANCELLATION POINT

int Condition::signal() {
 return ::pthread_cond_broadcast(&condition);


It doesn't matter that another package defines a class named Condition, because the fully-qualified name of my class is actually com::diag::grandote::Condition. Any source code that is placed within the namespace com::diag::grandote will automatically default to using my Condition, not the other library's Condition, eliminating the need for me to specify that long name every time.

Note too that the namespace incorporates the domain name of my company in reverse order - borrowing an idea from how Java conventionally organizations its source code and byte code files. This eliminates any collisions that might have occurred because another library itself has the same library, class, or module name as part of its own namespace.

This works great for C++, but the problem remains for C. So there, I resort to just prepending the library name and the module name to the beginning of every function. You already may have noticed that in the diminuto_mux_wait function in the Doxygen example above: the wait operation in the mux module of the diminuto library.

But one problem that neither C++ namespaces nor my C naming convention, solves is header file collisions. So in both C and C++ I organize header file directories in a hierarchical fashion so that #include statements look like this, from my Assay project.

#include <stdio.h>
#include <errno.h>
#include "assay.h"
#include "assay_parser.h"
#include "assay_fixup.h"
#include "assay_scanner.h"
#include "com/diag/assay/assay_scanner_annex.h"
#include "com/diag/assay/assay_parser_annex.h"
#include "com/diag/diminuto/diminuto_string.h"
#include "com/diag/diminuto/diminuto_containerof.h"
#include "com/diag/diminuto/diminuto_log.h"
#include "com/diag/diminuto/diminuto_dump.h"
#include "com/diag/diminuto/diminuto_escape.h"
#include "com/diag/diminuto/diminuto_fd.h"
#include "com/diag/diminuto/diminuto_heap.h"

The first two header files, stdio.h and errno.h, are system header files. The next four header files whose names begin with assay_ are private header files that are not part of the public API and which are in the source directory with the .c files being compiled. The next two header files are in a com/diag/assay subdirectory that is part of the Assay project. The remaining header files are in a com/diag/diminuto subdirectory that is part of the Diminuto project.

Here's another example, from the application gpstool that makes use of both my Hazer and Diminuto projects.

#include <assert.h>
#include <unistd.h>
#include <stdio.h>
#include <string.h>
#include <stdint.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <errno.h>
#include "com/diag/hazer/hazer.h"
#include "com/diag/diminuto/diminuto_serial.h"
#include "com/diag/diminuto/diminuto_ipc4.h"
#include "com/diag/diminuto/diminuto_ipc6.h"
#include "com/diag/diminuto/diminuto_phex.h"

This convention is obviously not so important in my C header files, where the project name is part of the header file name. But it becomes a whole lot more important in C++, where my convention is to name the header file after the name of the class it defines, as this snippet from Grandote illustrates.

#include <unistd.h>
#include <sys/stat.h>
#include <fcntl.h>
#include "com/diag/grandote/target.h"
#include "com/diag/grandote/string.h"
#include "com/diag/grandote/Platform.h"
#include "com/diag/grandote/Print.h"
#include "com/diag/grandote/DescriptorInput.h"
#include "com/diag/grandote/DescriptorOutput.h"
#include "com/diag/grandote/PathInput.h"
#include "com/diag/grandote/PathOutput.h"
#include "com/diag/grandote/ready.h"
#include "com/diag/grandote/errno.h"
#include "com/diag/grandote/Grandote.h"

Once I came upon this system of organizing header files, not only did it become much easier to integrate and use multiple projects into a single application, but the source code itself became more readable, because it was completely unambiguous where everything was coming from. This strategy worked so well, I use it not just for C and C++, but to organize my Python, Java, and other code too. I also organize my GitHub repository names, and, when I use Eclipse, my project names, in a similar fashion:

  • com-diag-assay,
  • com-diag-grandote,
  • com-diag-diminuto,
  • com-diag-hazer


This is more important than it sounds. Digital Aggregates Corporation ( is my consulting company that has been around since 1995. It holds the copyright on my open-source code that may find its way into my clients' products. The copyright of my closed-source code is held by another of my companies, Cranequin LLC ( So if I see the header file directory com/cranequin or a project or repository with the prefix com-cranequin, I am reminded that I am dealing with my own proprietary code under a different license.

Time (updated 2017-07-26)

The way time is handled in POSIX or Linux - whether you are talking about time of day, a duration of time, or a periodic event with an interval time - is a bit of a mess. Let's see what I mean.

gettimeofday(2), which returns the time of day, uses the timeval structure that has a resolution of microseconds. It's cousin time uses an integer that has a resolution of seconds.

clock_gettime(2), which when used with the CLOCK_MONOTONIC_RAW argument returns an elapsed time suitable for measuring duration, uses the timespec structure that has a resolution of nanoseconds.

setitimer(2), which is be used to invoke an interval timer, uses the itimerval structure that has a resolution of microseconds.

timer_settime(2), which also invokes an interval timer, uses the itimerspec structure that has a resolution of nanoseconds.

select(2), which is be used to multiplex input/output with a timeout, uses the timerval structure and has a resolution of microseconds. Its cousin pselect uses the timespec structure that has a resolution of nanoseconds.

poll(2), which can also be used to multiplex input/output with a timeout, uses an int argument that has a resolution of milliseconds. It's cousin ppoll uses the timespec structure that has a resolution of nanoseconds.

nanosleep(2), which is used to delay the execution of the caller, uses the timespec structure that has a resolution of nanoseconds. Its cousin sleep(3) uses an unsigned int argument that has a resolution of seconds. Its other cousin usleep(3) uses an useconds_t argument that is a 32-bit unsigned integer and which has a resolution of microseconds.

Seconds, milliseconds, microseconds, nanoseconds. So many different granularities of time. Several different types. It's too much for me.

Diminuto has two types in which time is stored: diminuto_ticks_t and diminuto_sticks_t. Both are 64-bit integers, the only difference being one is unsigned and the other is signed. (I occasionally regret even that.) All time is maintained in a single unit of measure: nanoseconds. This unit is generically referred to as a Diminuto tick.

diminuto_time_clock returns the time of day (a.k.a. wall clock time) in Diminuto ticks. There is another function in the time module to convert the time of day from a ticks since the POSIX epoch into year, month, day, hour, minute, second, and fraction of a second in - you guessed it - Diminuto ticks. The POSIX clock_gettime(2) API is used with the CLOCK_REALTIME argument.

diminuto_time_elapsed returns the elapsed time in Diminuto ticks suitable for measuring duration. It also uses the clock_gettime(2) API but with the CLOCK_MONOTONIC_RAW argument. This makes the returned values unaffected by changes in the system clock wrought by system administrators, by the Network Time Protocol daemon, by Daylight Saving Time, or by the addition of leap seconds. However, this means the value returned by the function is only really useful in comparison or in arithmetic operations with prior or successive values returned by the same function.

diminuto_timer_periodic invokes a periodic interval timer, and diminuto_timer_oneshot invokes an interval timer that fires exactly once, both using Diminuto ticks to specify the interval. Both are implemented using the POSIX timer_create(2) API with the CLOCK_MONOTONIC argument.

diminuto_mux_wait uses pselect(2) for I/O multiplexing, and diminuto_poll_wait does the same but uses ppoll(2), both specifying the timeout in Diminuto ticks. (I believe pselect is now implemented in the Linux kernel using ppoll, but that wasn't true when I first wrote this code and they had very different performance characteristics.)

diminuto_delay delays the caller for the specified number of Diminuto ticks. It is implemented using the POSIX nanosleep(2) API.

Floating Point (updated 2017-07-26)

If you really need to use floating point, you'll probably know it. But using floating point is problematic for lots of reasons. Diminuto avoids using floating point except in a few applications or unit tests where it is useful for reporting final results.

Even though Diminuto uses a single unit of time, that doesn't mean the underlying POSIX or Linux implementation can support all possible values in that unit. So every module in Diminuto that deals with time has an inline function that the application can call to find out what resolution the underlying implementation supports. And every one of those functions returns that value in a single unit of measure: Hertz. Hertz - cycles per second - used because it can be expressed as an integer. It's inverse is the smallest time interval supported by the underlying implementation.

diminuto_frequency returns 1,000,000,000 Hz, the base frequency used by the library. The inverse of this is one billionth of a second or one nanosecond.

diminuto_time_frequency returns 1,000,000,000 Hz, the inverse being the resolution of timespec in seconds.

diminuto_timer_frequency returns 1,000,000,000 Hz, the inverse being the resolution of itimerspec in seconds.

diminuto_delay_frequency returns 1,000,000,000 Hz, the inverse being the resolution of timespec in seconds.

With just some minor integer arithmetic - which can often be optimized out at compile-time - the application can determine what is the smallest number of ticks the underlying implementation can meaningfully support in each module. (Note that this is merely the resolution representable in the POSIX or Linux API; the kernel or even the underlying hardware may support an even coarser resolution.)


I first saw the countof technique in a VxWorks header file perhaps twenty years ago while doing embedded real-time development in C++ at Bell Labs. Similarly, a lot of these techniques have been picked up - or learned the hard way - while taking this long strange trip that has been my career. I have also benefitted greatly from having had a lot of mentors, people smarter than I am, who are so kind and generous with their time. Many of these techniques gestated in my C++ library Desperado, which I began putting together in 2005.

I started developing and testing the Diminuto C code for a number of reasons.
  • I found myself solving the same problems over and over again, sometimes for different clients, sometimes even for different projects for the same client. For one reason or another - sometimes good, sometimes not so good - the proprietary closed-source code I developed couldn't be shared between projects. But open-source code could easily be integrated into the code base.
  • I wanted to capture some of the design patterns I had found useful in my closed-source work. Working, unit tested, open-source code, developed in a clean room environment, and owned by my own company, was a useful way to do that.
  • I needed a way to get my head around some of the evolving C, C++, POSIX, and Linux APIs. Since I can only really learn by doing, I had to do, which for me meant writing and testing code.
  • I wanted to make an API that was more consistent, more easily interoperable, and less prone to error than was offered by raw POSIX, Linux, and GNU.
  • In the past I have been a big proponent of using C++ for embedded development. I have written hundreds of thousands of lines of C++ while doing such work over the years. But more recently, I have seen a gradual decline in the use of C++ by my clients, with a trend more towards segregating development into systems code in C and application code in languages like Python or Java. Although I miss C++, I have to agree with the economics that. And I fear that C++ has evolved into a language so complex that it is beyond the ken of developers that my clients can afford.
I am not a "language lawyer". I get paid to develop product that ships and generates revenue. I believe I have been successful in part because I am a very hands-on and evidence based person, but also because I am extremely pragmatic about what works and what doesn't.

This is some stuff that, over the span of many years, has worked.

Update (2017-04-14)

I recently forked Desperadito (which itself is a mashed up fork of both Desperado and Hayloft) into Grandote ( What's different is Grandote uses the Diminuto C library as its underlying platform abstraction. It requires that you install Diminuto, Lariat (a Google Test helper framework), and Google Test (or Google Mock). That's some effort, but at least for me it wasn't overly burdensome when I built it all this morning on a Ubuntu 16.04 system. The other improvement over Desperadito is that both the old Desperado hand coded unit tests, and the newer Hayloft Goggle Test unit tests, work. Grandote would be my C++ framework going forward, if I were to need my own C++ framework. (And Grandote doesn't preclude using STL or Boost as well.)