Back

TechnologySep 04, 2018

Developer’s Guide to Salesforce: DX and Source Control

Amanda Dunham and Brian Hafferkamp

the problem: source of truth and managing declarative changes

In Part 1 of this series, we discussed how Salesforce DX gives developers the ability to write custom Apex code locally and deploy it to a Sandbox or Production org, all using a command line interface. But as any good developer knows, the development process is not complete without a source of truth. It is all well and good to be able to make changes in your IDE of choice, but what if multiple developers are working on the same file at the same time? What if someone makes a change that later needs to be reverted? If there is no single place to track and manage changes, things will fall through the cracks, resulting in unnecessary work and frustration.

Additionally, while custom Apex code is important to a Salesforce environment, anyone who has clicked around in Salesforce knows that most of the functionality and customizations in an org exist outside of Apex classes. The power of Salesforce lies chiefly in the customization provided by the point-and-click tools, and it is just as necessary to track changes to these components as it is to track code. In the past, all of that work has been managed with Change Sets—the Salesforce solution for moving metadata changes between orgs. Change Sets require someone, likely a developer or an administrator, to manually track all files, flows, processes, objects, fields, etc., that have changed, and select each individual component to deploy to other environments. Keeping track of all these changes is tedious and very prone to error, and no single environment is guaranteed to be a source of truth.

the solution

Enter: the Scratch Org. Salesforce DX gives us the ability to create temporary, personal orgs where changes can be made in isolation, pulled into a local environment, entered into source control, and then deployed to Sandbox and Production orgs. This feature is made available when DevHub is enabled in an Enterprise, Performance, or Unlimited Edition org.

The Scratch Org can be thought of as a private development playground—a place where a developer can write code and customize configurations without the fear that their changes will interfere with anyone else’s work (or vice versa). They are created based on a JSON definition file that can be customized with a large set of configurations and preferences, allowing for a quick setup of an org that approximates the shape of a production org. Scratch Orgs are owned solely by the person who creates them, and they are simple to create and destroy. So simple, in fact, that Salesforce gives them a maximum lifespan of 30 days, a feature that prevents developers from working in a persistently stale environment. And though the Scratch Orgs are owned independently, the definition file can be managed in the repository so that all developers are working in identically shaped environments.

Scratch Orgs bring a whole new dimension to Salesforce development: Not only do they allow for code to be written in an IDE and pushed to a playground for testing, they also open up the possibility to capture the configuration changes made through the Salesforce UI—all of the “click, not code” work—in metadata xml files. This means that all declarative changes can be tracked in source control, something that was nearly impossible until now. While Salesforce has had the idea of a source of truth, most likely the Production org, it keeps no record of when, why, or by whom specific changes were made. These are essential features of source control for developers, so their absence makes it hard to consider any traditional Salesforce environment a true source of truth in the same way that a Git repository is.

With the introduction of Scratch Orgs and the ability to manage changes in a more traditional development environment, the role of Change Sets is heavily diminished. Customizations no longer need to be tracked manually; DX can capture all changes from developers’ Scratch Orgs in source control.

the drawbacks

While the Scratch Org has vastly improved the development life cycle within the Salesforce ecosystem, it is still a relatively new product that has potential for improvement.

One significant hurdle surrounding development in Scratch Orgs is the inability to create one with configurations that exactly mirror the DevHub it is created with. There are many settings that can be defined in the previously mentioned Scratch Org definition file, but it is tedious to figure out what settings you need. Many of the options are poorly documented, and some do not even work. Furthermore, there are some settings the definition file does not account for, so you are left with something that is close but not quite the same. These differences between the development and destination environments can cause unexpected issues when changes are moved from a Scratch Org to a Sandbox. They also make it difficult, if not impossible, to manage profiles with DX, which creates a lot of headaches and can leave behind a long list of manual updates to make after changes are deployed to a new org. As a solution for this, Salesforce has introduced the idea of Shapes, which would provide the ability to mirror Production in a Scratch Org. However, at the time of this writing, this is a feature that is still in pilot and not recommended for general use.

Another limitation presents itself when considering Salesforce’s recommendation to break an org into “artifacts,” and the fact that DX does not come with any sort of merge capabilities. The recommendation around artifacts is that the functionality and metadata in your org should be broken up into smaller pieces that can be managed independently. Salesforce also recommends that things like custom objects and fields should only be used in one artifact each—no sharing. However, this limitation is not realistic; when working with a highly customized org, it is likely that completely different workstreams will rely on the same metadata every once in a while. And if you do manage an object or a field in two different artifacts, whichever changes are most recent will always win, leaving you at risk of overwriting someone else’s work. Thankfully, Salesforce has also introduced a solution to this in second-generation packaging, a feature that, although still in Beta, has proven to be helpful in creating artifacts that contain shared metadata and can be listed as a dependency for other bodies of work. There is still room for improvement, but these new features prove that Salesforce is working hard to get this right; early adopters simply have to put up with the growing pains.

next time

We’ve now explored the possibilities offered by the command line interface and by Scratch Orgs, but even more can happen when we use them together. In the next installment of the series, we will walk through continuous integration and deployment (CI/CD) from source control using the DX CLI. We will also further expound on second-generation packaging and how it can be leveraged in the CI/CD process. Stay tuned!

Looking for help with Salesforce development? Reach out to us at findoutmore@credera.com. We are practitioners, not salespeople, so you’ll speak with an actual Credera consultant. We look forward to chatting to see how we might be able to help.

Have a Question?

Please complete the Captcha