Scenario Outline with 20 Examples rows, or a CSV dataset? What do reviewers expect here

Example threads. These questions and answers were written by the SDODS team to document real problems and how to fix them. The people named are illustrative, not real users. Questions asked on this site are marked new.

We have a shipping tax table: twenty countries, each with a rate and an expected label on the confirmation page. I wrote it as a Scenario Outline and the feature file is now a wall of pipes — the twenty rows are longer than the rest of the file put together, and two of the rates are different on staging than they are on production.

I could move it into a CSV. But then the feature file no longer says what the expected behaviour is, and the whole point of Gherkin was that a person can read it.

Where do people draw the line? Is there a row count where an Outline stops being the right thing?

data data-driven

asked by Jide Okonkwo

3 answers

It is not a row count, it is whether the table is documentation or input.
An Outline earns its place when the rows are the requirement — a handful of variants a reader needs to see to understand the behaviour. A dataset earns its place when the rows are many, or when they differ per environment. Yours does both: twenty rows nobody reads line by line, and two values that change with the environment. That settles it.
gherkin
```
Given I load dataset "shipping" where "country" is "DE"
When I check out with "{{firstName}}" "{{lastName}}" "{{postalCode}}"
Then I should see the text "{{expectedTax}}"
```
Then keep a small Outline for the two or three cases that document the behaviour — zero-rated, standard, and whatever the interesting edge is — and let the dataset carry the other seventeen as coverage.
The environment half is the part that does not survive an Outline at all. data/common/shipping.csv is the baseline and data/staging/shipping.csv overrides it, and the feature file does not have to know.
Data-driven design
answered by Priya VenkateshSDODS team
One thing to know before you move: an Outline reports each row as its own scenario, so a failure names the row. A dataset-driven scenario is one scenario, and a failure names the scenario.
That is a real loss, and the mitigation is to give the dataset a stable identifier column and put it somewhere the failure will show it — in the step arguments, or in the assertion text. If the only thing your report says is "checkout applies the right tax" and it failed, you are opening a trace to find out which country.
Outlines have a # title-format: comment for the same reason, so the generated scenario titles are readable.
answered by Hsu Wei-lin
Also tag it @data:shipping. It does not do anything at run time, but lint checks the dataset exists, and when somebody edits the CSV you can run everything that depends on it:
bash
```
sdods run -p shop -e staging -t "@data:shipping"
```
answered by Lena Hartwig

Know the answer?

Wrap commands and output in triple backticks and they will render as a code block.

Checking whether you’re signed in…

Have a different question? Ask it.

Scenario Outline with 20 Examples rows, or a CSV dataset? What do reviewers expect here

3 answers

Know the answer?

Related