I'm working with a process like this:
However there are some tests that are problematic, like
I would want to have a test for that - don't want any missing translations in production.
But I would not want to block the manual test if there's no Spanish translations - those translations can take time to come in, but won't change the test outcome.
Smaller stories may also be behind feature flags and thus could be merged without being polished (or having translations).
How do you deal with this? Do you have some tests (like missing translations) that don't block the build, and others that do?
Ben Buchanan (200ok)
I make some bits of the web.
Many(/most?) test frameworks have some kind of flag system build in to disable some tests/suites based on an environmental variable. So you can enable or disable when
envis set toci. Have done it from time to time.Generally I prefer tests to run pre-commit during dev; then again when the PR is raised (blocking merge if they fail); then again after merge to master. Releases are done off master with a check for a green build.