Tweet

SalesforceDX Week 1

(NOTE: This post is based on the SalesforceDX pilot which, like all pilots, may never make it to GA. I bet it does though!)

Introduction

The SalesforceDX pilot started a week or two ago and BrightGen were lucky enough to be selected to participate (thanks to the sterling efforts of my colleague Kieran Maguire who didn’t screw up his signup, unlike me!). This week I’ve managed to spend a reasonable amount of time reading the docs and trying out the basics and it’s clear already that this is going to be a game changer. 

There will be bugs!

While this isn’t the first  pilot that I’ve been involved in, but it’s by far the largest in terms of new functionality - a new version of the IDE, a new CLI with a ton of commands and a new type of org. A pilot is a two way street - you get to play with the new feature long before it becomes (if it ever does!) GA, but the flip side is that this won’t be tested to destruction like a GA feature. With the best will in the world there’s no way that Wade Wagner and co could test out every possible scenario, so some stuff will break, and that’s okay. When things break (or work in a non-intuitive way) you sometimes get a chance to influence how the fix works, which is pretty cool. Be a grown up though - report potential issues in a measured way with as much detail as you can gather - it’s always embarrassing when you have to climb down from a high horse when you realise that you made the mistake, not the tool!

Scratch Orgs

Scratch orgs are probably the feature I’ve been most excited about in SalesforceDX. I run the BrightMEDIA team at BrightGen and setting up a developer org for a new member of our team takes around half a day. After the initial setup, every release needs to be executed on each dev org as well as the target customer or demo org(s), which consumes a fair amount of time with a weekly release cadence. There’s also the problem of experimentation - often devs will try something out, realise it’s not the best way to do it, but not tear down everything they built. Over time the dev org picks up baggage which the dev has to be careful doesn’t make its way into version control.

Scratch orgs mitigate the first problem and solve the second. A scratch org is ephemeral - it is created quickly from configuration and should only last for the duration of the development task you are carrying out. When we setup a developer edition we have to contact Salesforce support to get the apex character limit increased and multi-currency enabled. Scratch orgs already have an increased character limit and features can be defined in the configuration. Here’s the scratch org configuration file for one of my projects:

{
  "Company": "KAB DEV",
  "Country": "GB",
  "LastName": "kbowden",
  "Email": "keir.bowden@googlemail.com",
  "Edition": "Developer",
  "Features": "Communities;MultiCurrency",
  "OrgPreferences" : {
    "ChatterEnabled": true,
    "S1DesktopEnabled" : true,
    "NetworksEnabled": true,
    "Translation" : true,
        "PathAssistantsEnabled" : true
  }
}

The features attribute : 

"Features": "Communities;MultiCurrency"

enables communities and multi-currency when my org is created, saving me a couple of hours raising a case and waiting for a response right off the bat.

Creating a Scratch Org

Is a single command utilising the new CLI:

> sfdx force:org:create --definitionfile config/workspace-scratch-def.json

 and it’s fast. I’ve just created an org for the purposes of this blog and I’d be surprised if it took more than a minute, although the DNS propagation of the new org name can take a few more minutes. You don’t have to worry about passwords with scratch orgs, it’s all handled by the CLI. To “login” I just execute:

> sfdx force:org:open

and a browser window opens and I’m good to go. Accessing the Manage Currencies setup node shows that multi-currency has indeed been enabled.

There’s a bit more to it than this in our case - a few packages have to be installed for example - but so far it looks like I can script all of this, which means a new developer just runs a single command to get an org they can start work in. Note that there’s just the standard developer edition data in here - I haven’t found time to play with the data export/import side of the CLI yet so that will have to wait for another day.

Managing Code

If you are familiar with the git paradigm of pulling and pushing changes from/to a remote location, the SalesforceDX source management is simple to pick up. You don’t get version control, but you do get automatic detection of what has changed and where. The docs state that this functionality is only available for scratch orgs and we still have to use the metadata API to push to sandbox/production orgs, which seems fair enough a pilot to me.

Detecting Differences

In my scratch org I create a simple lightning component in the developer console:

<aura:component >
	<h1>I'm a simple Lightning Component</h1>
</aura:component>

In current development process I have a script to extract the Lightning metadata and copy it into my source directory. With scratch orgs it’s a fair bit easier.

I can figure out what has changed by running the status subcommand:

> sfdx force:source:status

State Full Name Type Workspace Path
────────── ───────── ──────────────

Remote Add Simple AuraDefinitionBundle

Pulling Code from the Scratch Org

 to extract the new code from the org to my workspace in local filesystem:

> sfdx force:source:pull

State   Full Name Type                 Workspace Path
─────── ───────── ──────────────────── ───────────────────────────────────────────────────────────
Changed Simple    AuraDefinitionBundle /Users/kbowden/SFDX/Blog/force-app/main/default/aura/Simple

I can then list the contents of my workspace and there is my new component:

> ls force-app/main/default/aura/

Simple

Pushing Code to the Scratch Org

If I edit the component locally, the status subcommand picks that up too:

> sfdx force:source:status

State         Full Name         Type                 Workspace Path
───────────── ───────────────── ──────────────────── ──────────────────────────────────────────────────────
Local Changed Simple/Simple.cmp AuraDefinitionBundle force-app/main/default/aura/Simple/Simple.cmp-meta.xml
Local Changed Simple/Simple.cmp AuraDefinitionBundle force-app/main/default/aura/Simple/Simple.cmp

and I can publish these changes to the scratch org via the push subcommand:

> sfdx force:source:push

State   Full Name Type                 Workspace Path
─────── ───────── ──────────────────── ──────────────────────────────────
Changed Simple    AuraDefinitionBundle force-app/main/default/aura/Simple

Scratch Orgs are Temporary

Unlike developer orgs, scratch orgs are not intended to persist. In fact I’ve seen docs that state they may be deleted at any point in time. Although I’d imagine in reality it will be based on lack of use, it doesn’t matter as if your scratch org disappears, you can just spin up a new one with the same setup, push your local code and you are back where you were. This does mean you need to treat your local filesystem as the source of the truth, but that’s pretty much how I work anyway.

This way scratch orgs don’t accumulate any baggage, and you don’t have to worry about destroying anything. If you don’t put it into version control, it won’t be there in the future.

One Org Shape to Rule them All

The configuration, data and code that make up your scratch org can be considered a template, especially if the setup is all scripted. This means that my team and I just need to update a single org shape “template" with changes that need to be applied to every development environment. Then we just spin up new scratch orgs and we can be sure that we are all in step with each other, which will save us time on many levels. 

Related Posts

• SalesforceDX - View from the coal face
• SalesforceDX official site

 Follow @bob_buzzard