[bartek]

The GCS fetcher, emitter and pipes iterator all fail to build without a projectId, but none of their calls need one. They read, write and list objects in a specified bucket.

The java client also does not require projectId, and a comment within it states "Project ID is only required for creating buckets, so we don't requite it for creating the service.":

https://github.com/googleapis/google-cloud-java/blob/123f97c3f843dc7a3ff7b4ef0596c1425502a275/java-storage/google-cloud-storage/src/main/java/com/google/cloud/storage/StorageOptions.java#L326

The docs also state projectId picks the project billed for API calls, but storage access is billed to the project that owns the bucket; this has been corrected.

[tballison]

nits from 🤖

 - No CHANGES.txt row. This relaxes a user-facing config requirement, so it probably deserves one under 4.2.1.
  - Test coverage is uneven. The fetcher gets with-project, without-project and requires-bucket tests. The emitter and iterator only get without-project.
  - Examples still set "projectId": "my-gcp-project" in all four docs/modules/ROOT/examples/pipes-gcs-*.json files. Dropping it from at least one would show
    it's optional.

Only thing that matters to me is the CHANGES.txt. Examples are nice to have, but not necessary. Test coverage... whatever.

[bartek]

@tballison Thanks! I've pushed a fixup. I'll get more familiar with requirements regarding CHANGES.txt so I don't miss that again.