AdaCore Blog

Witnessing the Emergence of a New Ada Era

by Quentin Ochem

For nearly four decades the Ada language (in all versions of the standard) has been helping developers meet the most stringent reliability, safety and security requirements in the embedded market. As such, Ada has become an entrenched player in its historic A&D niche, where its technical advantages are recognized and well understood. Ada has also seen usage in other domains (such as medical and transportation) but its penetration has progressed at a somewhat slower pace. In these other markets Ada stands in particular contrast with the C language, which, although suffering from extremely well known and documented flaws, remains a strong and seldom questioned default choice. Or at least, when it’s not the choice, C is still the starting point (a gateway drug?) for alternatives such as C++ or Java, which in the end still lack the software engineering benefits that Ada embodies..

Throughout AdaCore’s twenty-five year history, we’ve seen underground activities of software engineers willing to question the status quo and embark on new technological grounds. But driving such a change is a tough sell. While the merits of the language are usually relatively easy to establish, overcoming the surrounding inertia often feels like an insurmountable obstacle. Other engineers have to be willing to change old habits. Management has to be willing to invest in new technology. All have to agree on the need for safer, more secure and more reliable software. Even if we’ve been able to report some successes over the years, we were falling short of the critical mass.

Or so it seemed.

The tide has turned. 2018 and 2019 have been exceptional vintages in terms of Ada and SPARK adoption, all the signs are showing that 2020 will be at least as exciting. What’s more - the new adopters are coming from industries that were never part of the initial Ada and SPARK user base. What used to be inertia is now momentum. Let’s take a look at the information that can be gathered from the web over the past two years to demonstrate the new dynamic of Ada and SPARK usage.

The Established User Base

Before talking about new adopters, it’s important to step back and re-establish the basis of the Ada and SPARK usage, which is the root of its viability over the very long term. Ada and SPARK are used by a very large user base in the defense and avionics domains. A glance at AdaCore customer list - a subset of the actual user base - will give a good idea of the breadth of technology usage. A lot of the projects here have lifetimes over decades, some started in the early days of Ada in the mid 80’s carried all the way to the present, some have already planned lifetimes spanning over the next two decades. Projects range from massive air traffic management systems running on vast arrays of servers to embedded controllers running on aircraft engines, sensors, or satellite flight control systems with extremely stringent resource constraints. Some applications are still maintained today on hardware dating as far back as Motorola 68K or Intel i386 series, while others are deployed on the latest ARM Cortex or RISC-V cores. Most have some level of reliability constraints, up to the highest levels of the avionics DO-178B/C standard. 

Due to the nature of the domain, it is difficult to communicate specifically about these projects, and we only have scarce news. One measure of the increasing interest in Ada and SPARK can be inferred from defense-driven research projects which contain references to these language technologies. The most notable example is the recent UK-funded HICLASS project, focused on security, which involves a large portion of the UK defense industry. Some press releases are also available, in particular in the space domain (European Space Agency, AVIO and MDA). These data samples are representative of a very active and vibrant community which is committed to Ada and SPARK for decades to come - effectively guaranteeing their industrial future as far as we can reasonably guess.

The Emerging Adopters

The so-called “established user base” has fueled the Ada and SPARK community up until roughly the mid 2010s. At that point of time, a new trend started to emerge, from users and use cases that we’had never seen before. While each case is a story in its own right, some common patterns have emerged. The starting point is almost always either the increase of safety or security requirements, or a wish to reduce the costs of development of an application with some kind of high reliability needs. This is connected to the acknowledgement that the programming language in use - almost exclusively C or C++ - may not be the optimal language to reach these goals. This is well documented in the industry; C and C++ flaws have been the subject of countless papers, and the source of catastrophic vulnerability exploits and tools to work around issues. The technical merits of Ada and its ability to prevent many of these issues is also well documented - we even have access to some measurements of the effects. The most recent one is an independent study developed by VDC, which measured up to 38% cost-savings on Ada vs C in the context of high-integrity markets that have adopted Ada for a long time.

We’re talking a lot about Ada here, but in fact new adopters are typically driven by a mix of SPARK and Ada. The promise that SPARK offers is automatic verification of software properties such as absence of buffer overflow, together with stringent mitigation of others - and this by design, early in the development process. This means that developers are able to self-check their code - not only is the code more reliable, it is also more reliable straight out as you write it, avoiding many mistakes that could otherwise pass through testing, integration or deployment phases.

Some of the SPARK adopters motivated by these benefits come from academia. Over the past 2 years, over 40 universities have joined the GNAT Academic Program (“GAP”), with a mix of teaching and research activities, including for example FH Campus Wien train project, CubeSat and UPMSat-2.

Many adopters can also be found in industry. Some of the following references highlight teams at the research phase, some others represent projects already deployed. They all however contribute to this solid wave of new Ada and SPARK adopters. The publications referenced in the following paragraphs have been published between 2018 and 2019.

One obvious application for Ada and SPARK, where human lives are at risk, is the medical device domain. So it comes without surprise that this area is amongst those adopting the technology. Two interesting cases come to mind. The first one in RealHeart, a Scandinavian manufacturer that is developing an artificial heart with on-board software written in Ada and SPARK, who issued a press release and later made an in-depth presentation at SPARK & Frama-C days. The second reference comes from a large medical device corporation, Hillrom, who published a paper explaining the rationale for the selection of SPARK and Ada for development of ECG algorithms.

Another domain is everything that relates to security. The French security agency ANSSI studied various languages to implement a secure USB key and selected SPARK as the best choice. They published a research paper, presentation and source code. Another interesting new application has been implemented by a German company Componolit developing proven communication protocols

Of course, established markets are also at the party. The University of Colorado’s Laboratory for Atmospheric and Space Physics has recently adopted Ada to develop an application for the International Space Station. In the defense domain, the Air Force Research Labs is studying re-writing a drone framework from C++ to SPARK and doing functional proofs, with a public research paper and source code available.

While all of these domains provide interesting adopter stories, the one single domain that has demonstrated the most interest in the recent past is undoubtedly automotive. This is probably coming from the increasing complexity of electronics systems in cars, with applications such as Advanced Driver Assistance Systems (ADAS) and autonomous vehicles. References in this domain ranges from tier 1 suppliers such as Denso or JTEKT as well as OEMs and autonomous vehicle companies like Volvo’s subsidiary Zenuity

And there’s NVIDIA.

In January of this year, we published with NVIDIA a press release and a blog post, followed-up this November by a presentation at our annual Tech Days conference, and an on-line webex (also see the slides for the webex). In many respects, this is a unique tipping point in the history of Ada adoption in terms of impact in a non-A&D domain, touching considerations ranging from security to automotive safety, all under the tight constraints of firmware development. The webex in particular provides a unique dive into the reasons behind the adoption of SPARK and Ada by a company that didn’t have any particular ties to it initially. It also gives key insights on the challenges and costs of such an adoption, together with the benefits already observed. In many respects, this is almost an adoption guide to the technology from a business standpoint.

Wrapping Up

Keep in mind that the above references are only those that are publicly available, which we know about. There are many more projects under the hood, and even more that we’re not even aware of. Everything considered, this is a very exciting time for the Ada and SPARK languages. Stay tuned, we have an array of new stories coming up for the months and years to come!

Posted in

About Quentin Ochem

Quentin Ochem has a software engineering background, specialized in software development for critical applications. He has over 10 years of experience in Ada development. His responsibilities at AdaCore include leading technical account management as well as driving business development, following projects related to avionics, railroad, space and defense industries.