Simplicity is a form of art...

Creating an enterprise open source policy
by Sven Vermeulen, post on Sat 20 November 2021

Nowadays it is impossible to ignore, or even prevent open source from being active within the enterprise world. Even if a company only wants to use commercially backed solutions, many - if not most - of these are built with, and are using open source software.

However, open source is more than just a code sourcing possibility. By having a good statement within the company on how it wants to deal with open source, what it wants to support, etc. engineers and developers can have a better understanding of what they can do to support their business further.

In many cases, companies will draft up an open source policy, and in this post I want to share some practices I've learned on how to draft such a policy.

Assess the current situation

When drafting a policy, make sure you know what the current situation already is. Especially when the policy might be very restrictive, you might be facing a huge backlash from the organization if the policy is not reflecting the reality. If that is the case, and the policy still needs to go through, proper communication and grooming will be needed (and of course, the "upper management hammer" can help out as well).

Often, higher management is not aware of the current situation either. They might think that open source is hardly in use. Presenting them with facts and figures not only makes it more understandable, it will also support the need for a decent open source policy.

When you have a good view on the current usage, you can use that to track where you want to go to. For instance, if your company wants to adopt open source more actively, and pursue open source contributions, you might want to report on the currently detected contributions, and use that for follow-up later.

Get HR and compliance involved

Before you embark on the journey of developing a decent open source policy, make sure you have management support on this, as well as people from HR and your compliance department (unless your policy will be extremely restrictive, but let's hope that is not the case).

You will need (legal &) compliance involved in order to draft and assess the impact of internal developers and engineers working on open source projects, as well as the same people working on open source projects in their free time. Both are different use cases but have to be assessed regardless.

HR is generally involved at a later stage, so they know how the company wants to deal with open source development. This could be useful for recruitment, but also for HR to understand what the policy is about in case of issues.

An important consideration to assess is how the company, and the contractual obligations that the employees have, deals with intellectual property. In some companies, the contract allows for the employees to retain the intellectual property rights for their creations outside of company projects. However, that is not always the case, and in certain sectors intellectual property might be assumed to be owned by the company whenever the creation is something in which the company is active. And that might be considered very broadly (such as anything IT related for employees of an IT company).

The open source policy that you develop should know what the contractual stipulations say, and clarify for engineers and developers how the company considers the intellectual property ownership. This is important, as it defines who can decide to contribute something to open source.

Understand and simplify license requirements

Many of the decisions that the open source policy has to clarify will be related to the open source licenses in use. Moreover, it might even be relevant to define what open source is to begin with.

A good source to use is the Open Source Definition as published and maintained by the Open Source Initiative (OSI). Another definition is the one by the Free Software Foundation titled "What is free software and why is it so important for society".

The license is the agreement that the owner of the software puts out that declares how users can use that software. Most, if not all software that a company uses, will have a license - open source or not. But most commercial software titles have specific licenses that you need to go through for each specific product, as the licenses are not reused. In the open source world, licenses are reused so that end users do not need to go through product-specific terms.

The OSI organization has a list of approved licenses. However, even amongst these licenses, you will find different types of licenses out there. While they are commonly grouped into copyleft and permissive open source licenses, there are two main categories within the copyleft licenses that you need to understand:

  • strong copyleft licenses that require making all source code available upon distribution, or sometimes even disclosure of the application base
  • "scoped" copyleft licenses that require making only the source code available of the modules or libraries that use the open source license (especially if you modified them) without impacting the entire application

While the term "strong copyleft" is something that I think is somewhat generally accepted (such as in the Snyk article "Open Source Licenses: Types and Comparison" or in Wikipedia's article), I do not like to use its opposite "weak" term, as the licenses themselves do not reduce the open source identity from the code. Instead, they make sure the scope of the license is towards a particular base (such as a library) and not the complete application that uses the license.

Hence, open source policies might want to focus on those three license types for each of the use cases:

  • permissive licenses, like Apache License 2.0 or MIT
  • scoped copyleft licenses, like LGPL or EPL-2.0
  • strong copyleft licenses, like GPL or AGPL

Differentiate on the different open source use cases

There are several use cases that the policy will need to tackle. These are, in no particular order:

  • Using off-the-shelf, ready-to-use open source products
  • Using off-the-shelf libraries and modules for development
  • Using open source code
  • Contributing to open source projects for company purposes
  • Contributing to open source projects for personal/private purposes
  • Launching and maintaining open source projects from the company

Each of these use cases might have their specific focuses. Combine that with the license categories listed earlier, and you can start assessing how to deal with these situations.

For instance, you might want to have a policy that generally boils down to the following:

  • When using off-the-shelf, ready-to-use open source products, all types of products are allowed, assuming the organization remains able to support the technologies adopted. Furthermore, the products have to be known by the inventory and asset tooling used by the company.
  • When using libraries or modules in development projects, only open source products with permissive or scoped copyleft licenses can be used. Furthermore, the libraries or modules have to be well managed (kept up-to-date) and known by the inventory and asset tooling used by the company.
  • When using open source code, only code that is published with a permissive license can be used. At all times, a reference towards the original author has to be retained.
  • When contributing to open source projects for company purposes, approval has to be given by the hierarchical manager of the team. Contributions have to be tagged appropriately as originating from the company (e.g. using the company e-mail address as author). Furthermore, employees are not allowed to contribute code or intellectual property that is deemed a competitive advantage for the company.
  • When contributing to open source projects for personal/private purposes, employees are prohibited to use code from the company or to do contributions using their company's e-mail address. However, the company does not claim ownership on the contributions an employee does outside the company's projects and hours.
  • When creating new projects or publishing internal projects as open source, sufficient support for the project has to be granted from the company, and the publications are preferentially done within the same development services (like version control) under management of the company. This ensures consistency and control over the company's assets and liability. Projects have to use a permissive license (and perhaps even a single, particular license).

Or, if the company actively pursues an open source first strategy:

  • Off-the-shelf, ready-to-use open source products are preferred over propriatary products. Internal support teams must be able to deal with general maintenance and updates. The use of commercially backed products is not mandatory, but might help when there is a need for acquiring short-term support (such as through independent consultants).
  • Development projects must use projects that use permissive or scoped copyleft licenses for the libraries and dependencies of that project. Only when the development project itself uses a strong copyleft license are dependencies with (the same) strong copyleft license allowed. Approval to use a strong copyleft license is left to the management board.
  • Engineers and developers retain full intellectual property rights to their contributions. However, a Contributor License Agreement (CLA) is used to grant the company the rights to use and distribute the contributions under the license mentioned, as well as initiate or participate in legal actions related to the contributed code.

Clarify what is allowed to be contributed and what not

In the above example I already indicated a "do not contribute code that is deemed a competitive advantage" statement. While it would be common sense, companies will need to clarify this (if they follow this principle) in their policies so they are not liable for problems later on.

A competitive advantage primarily focuses on a company's crown jewels, but can be extended with code or other intellectual property (like architectural information, documentation, etc.) that refers to indirect advantageous solutions. If a company is a strong data-driven company that gains massive insights from data, it might refuse to share its artificial intelligence related code.

There are other principles that might decide if code is contributed or not. For instance, the company might only want to contribute code that has received all the checks and controls to ensure it is secure, it is effective and efficient, and is understandable and well-written. After all, when such contributions are made in name of the company, the quality of that code reflects upon the company as well.

I greatly suggest to include examples in the open source policy to clarify or support certain statements.

Assess the maturity of an open source product

When supporting the use of open source products, the policy will also have to decide which open source products can be used and which ones can't. Now, it is it possible to create an exhaustive list (as that would defeat the purpose of an open source policy). Instead, I recommend to clarify how stakeholders can assess if an open source product can be used or not.

Personally, I consider this from a "maturity" point of view. Open source products that are mature are less likely to become a liability within a larger company, whereas products that only have a single maintained (like my own cvechecker project) are not to be used without understanding the consequences.

So, what is a mature open source project? There are online resources that could help you out (like the Qualipso-originated Open Source Maturity Model (OSMM)), but personally I tend to look at the following principles:

  • The project has an active development, with more than 5 active contributors in the last three months.
  • The project is visibly used by several other projects or products.
  • The project has well-maintained documentation, both for developers and for users. This can very well be a decent wiki site.
  • The project has an active support community, with not only an issue system, but also interactive services like forums, IRC, Slack, Discord, etc.
  • The project supports more than one major version in parallel, and has a clear lifecycle for its support (such as "major version is supported up to at least 1 year after the next major version is released").
  • The project publishes its artefacts in a controlled and secure manner.

A policy is just the beginning, not the end

As always, there will be situations where a company wants to allow a one-off case to deviate from the policy. Hence, make clear how deviations can be targeted.

For instance, you might want to position an architecture review board to support deviations from the license usage. When you do, make sure that this governance body knows how to deal with such deviations - understanding what licenses are, what the impact might be towards the organization, etc.

Furthermore, once the policy is ready to be made available, make sure you have support for that policy in the organization, as well as supporting tools and processes.

You might want to include an internal community to support open source/free software endeavors. This community can help other stakeholders with the assessment of a product's maturity, or with the license identification.

You might want to make sure you can track license usage in projects and deployments. For software development projects, there are plenty of commercial and free services that scan and present license usage (and other details) for a project. Inventory and asset management utilities often also include identification of detected software. Validate that you can report on open source usage if the demand comes up, and that you can support development and engineering teams in ensuring open source usage is in line with the company's expectations.

The company might want to dedicate resources in additional leakage detection and prevention measures for the open source contributions. While the company might already have code scanning techniques in place in their on-premise version control system, it might be interesting to extend this service to the public services (like GitHub and GitLab). And with that, I do not want to imply using the same tools and integrations, but more on a functional level.

Finishing off

A few companies, and most governmental organizations, publish their open source policies online. The TODO Group has graceously drafted a list of examples and templates to use. They might be a good resource to use when drafting up your own.

Having a clear and understandable open source policy simplifies discussions, and with the appropriate support within the organization it might jumpstart initiatives even further. Assuming the policy is sufficiently supportive of open source, having it published might eliminate the fear of engineers and developers to suggest certain open source projects.

Feedback? Comments? Don't hesitate to drop me an email, or join the discussion on Twitter.