Code review comment for lp:~elopio/snapcraft/integration_coverage

> I'm fine with the approach of a second, coverage-enabled binary. But don't we
> want to combine the unit tests and integration tests for the coverage report?

I've pushed a small change that combines the two reports.

I don't like this, but for now seems ok to me.

What I would prefer is to do test-driven development. So the unit test coverage should be close to 100%, and the unit coverage report will be just a guide of how well we are doing TDD.
On top of that, I would use the functional tests to cover the happy paths of the user-facing features. Plus some regressions tests or tests for important error conditions. Then the integration coverage report won't be close to 100%, and the simple table is not useful. What's useful is the HTML report that we can use to discover important user-facing features not covered.

There is no correct approach here, so we can discuss and feel free to disagree. If we decide to go the full TDD way, we can split the reports later and display the HTML nicely in the jenkins server.

thanks for the reviews!