Each of the five tests done so far in the climate-api took a single country code.
Change: For direct mode tests (as well as Servirtium record and playback modes), add the possibility of calculating rainfall averages for more than one country code (changes in our test harness). That would be a list of countries. Here's the Python test for a new test for 'gbr' and 'fra' rainfall average (pseudo code again):
test_averageRainfallForGreatBritainAndFranceFrom1980to1999CanBeCalculatedFromTwoRequests()
assert climateApi.getAveAnnualRainfall(1980, 1999, "gbr", "fra") == 951.3220963726872
Yes, the 'gbr+fra' test hits the WordBank climate web-api twice. Yes, that breaks the facade pattern, but this is just a test harness for Servirtium's per-language development.
In the Markdown servirtix grammar, there's an interaction number that shows the order of the TWO interactions (one for 'gbr' and one for 'fra')
## Interaction 0: GET /climateweb/rest/v1/country/annualavg/pr/1980/1999/gbr.xml
... markdown sections for: request headers, request body, response headers, response body ...
## Interaction 1: GET /climateweb/rest/v1/country/annualavg/pr/1980/1999/fra.xml
... markdown sections for: request headers, request body, response headers, response body ...
Most likely your well-factored code makes it easy to record a series of interactions.
There now should be 18 tests: 6 direct, 6 for record mode, and 6 for playback. Shared code should be a reality - ideally all the tests are written once and used three times (direct, record, and playback modes).