As development teams strive to deploy changes quickly and frequently, using a compare tool as part of your deployment pipeline, can help to ensure that item’s changes are propagated correctly across different environments. By identifying differences accurately, you could deploy changes with confidence and reduce the risk of errors and inconsistencies.
Today, with the addition of Change review, Deployment pipeline offers a complete built-in compare tool: It constantly runs an automated synchronization check, allows you to run the essential action of Compare to spot the New, Removed, and Different items when two not-in-sync environments are compared, and now, with Change review, you can also view the changes, line-by-line, in the item’s code.
In Change review, the changes of supported items are presented in a dedicated pop-up window, designed by industry standards for code compare. To access this window, select the Change review button next to the ‘Different’ label (appears once you hover the item).
At the Change review window, the compared content appears side-by-side by default, while the to-be-modified after deployment stage appears on the left, and the updated stage you’re about to deploy appears on the right (opposite to their direction in the pipeline page):
For your convenience, you can choose to switch the view to an inline view, any time. Your preference will be kept for the next time you’ll visit this page.
A navigation between the changed lines can be easily done by using the next/previous arrows, jump between the changes marked on the navigation bar or by scrolling freely.
Currently, only Dataset and Dataflow are supported and eligible for the change review feature. Paginated report is underway.
Change review is already available today so give it a try! Learn more.
Deployment enhancement –
Don’t miss deploying the rest of the items when one of them failed
We keep working on enhancing your deployment experience!
While until today, by default, a deployment stopped when failed to deploy one of its items so the subsequent items were not deployed too, now you have the option to select continuing the deployment so the downstream items of the failed item will be skipped, and still other subsequent items will be deployed.
As an example, let’s say I deploy 5 items, in this order, which are related in the following way:
- Dataset A -> report1, report2
- Dataset B
- Dataset C
If the box is checked and the deployment of Dataset A fails, the deployment will skip report1 and report2, and continue to deploy Dataset B and Dataset C.
Otherwise (if the box is not checked), non will be deployed.
Learn more about this option here.