Design renamed "Continue" to "Next" on the checkout stepper. Nothing failed. The nightly regression run is green and carries forty-one heal events, all with the same description, all healing from the accessible name to the test id.
Green is not the reaction I expected, and I am not sure what the right move is. Do I update one page object and hope? Do I go through forty scenarios?
Accepted answer
One page object, almost certainly. Forty-one events with one description means one declared locator used by forty-one scenarios, not forty-one problems. That is the shape healing is for — the blast radius of a rename is a single edit, and the run stayed green long enough for you to find out what the edit is.
sdods heal report --lastThe suggested column on that row is the selector that won every time. Make it the primary, move Continue out of the context and Next in, and re-run. If the count goes to zero you are done.
The case to actually worry about is the opposite one: forty-one events across forty-one different descriptions. That is a design system change, and it wants the reviewer agent rather than an afternoon of typing.
Before you edit, check that the test id it healed to is the element you mean. A stepper usually has more than one forward control — a primary button and a keyboard-accessible link, say.
A test id winning at its full base score tells you it matched exactly one visible, enabled element, which is reassuring, but it does not tell you the product team did not add a second "Next" somewhere. pageUrl on the event narrows it down quickly.
For the "now what" beyond this one rename: this is what the fragility number exists for. Once a locator has healed three or more times and its fragility clears the threshold it is marked hot, and insights ranks it next to the flaky scenarios.
That needs a database with ingested runs, so sdods run --ingest and then sdods insights compute -p <slug>. Without one you still have sdods heal report --all, which gets you the same ranking by hand.
Small process thing that helped us. After fixing the page object, run the affected feature with --repeat-each 10 --fail-on-flaky before you close the ticket. A rename tends to arrive with a timing change attached, and you would rather find that now than in the next nightly.
Have the same question?
Ask it with the command you ran and the output you got, and it will be answered here.
Ask a question