Gherkin testdata duplication

[aurelien-reeves]

🤔 What's the problem you've observed?

Testdata for gherkin are duplicated into each implementation. In case of update of the testdata, it generates huge diffs which are making PRs difficult to review properly (cf. cucumber/common#1741)

✨ Do you have a proposal for making it better?

Stop duplicating testdata and refer to those using relative paths
Or at least duplicating the testdata during the build, but ignoring those from the repo

📚 Any additional context?

Why the purpose of the duplication? Is it because of the subrepos?

---

This text was originally generated from a template, then edited by hand. You can modify the template here.

[mpkorstanje]

Why the purpose of the duplication? Is it because of the subrepos?

Yup. The subrepo sync would copy the language subtree. If the data wasn't duplicated, the sub repos build would fail.

[mpkorstanje]

I do think that fixing this is tied up with extracting gherkin from the common repo. I believe that PHP still requires a subrepo so we may not be able to get rid of it so easily. On the otherhand, if we can make php work withouth a subrepo, we can do the deduplication.

[ciaranmcnulty]

PHP doesn't need test data in the subrepo, I don't think

[aurelien-reeves]

Why the purpose of the duplication? Is it because of the subrepos?

Yup. The subrepo sync would copy the language subtree. If the data wasn't duplicated, the sub repos build would fail.

There are no automated build within subrepos. And I am not sure any subrepo has already been used as a dev environment and thus needed to actually build

I might submit a PR to remove that duplication in the monorepo. Any objection?

[mpkorstanje]

Partially fixed with #3

[mpkorstanje]

Good enough for now. Dart is it's own problem.