docs: sync env var docs with what the code reads#96
Conversation
Make README, .env.example, and the code agree: every variable the code reads via os.environ is documented and has a placeholder, and nothing unread appears in either. - fix cp .env-example -> .env.example (broke the first quickstart step) - drop DEFAULT_MODEL/MAX_TOKENS/TEMPERATURE from README (no code reads them) - drop OPENALEX_API_KEY/UNPAYWALL_EMAIL from .env.example (no code reads them) - document CONTACT_EMAIL and CLOUD_ML_REGION - add the NMDC_* ingestion-tuning vars to README and .env.example - fix a non-existent test path in the README Closes #95 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
There was a problem hiding this comment.
Pull request overview
Updates project documentation and the sample .env file so they reflect the environment variables the Python code actually reads (via os.environ.get(...)), and fixes a couple of quickstart/testing instructions that were pointing at the wrong filenames.
Changes:
- Fix Quick Start copy command to use
.env.example. - Update README configuration docs to remove unused variables and document the env vars that are actually read (including ingestion-tuning variables and defaults).
- Update
.env.exampleto include the ingestion-tuning variables and remove placeholders for env vars not read by the code.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| README.md | Aligns documented env vars and examples with current runtime behavior and existing test files. |
| .env.example | Adds placeholders for env vars read by code (ingestion/HTTP tuning) and removes ones that aren’t used. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
CLOUD_ML_REGION and GEMINI_REGION shipped as the literal placeholder 'region'. GEMINI_REGION is read first and only falls back to CLOUD_ML_REGION when unset, so a copied .env with GEMINI_REGION=region sends 'region' to Vertex and gets a confusing 404. Ship the real default (us-east5) and correct the comment (it takes precedence when set, it does not default to CLOUD_ML_REGION unless unset). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
| - `GOOGLE_APPLICATION_CREDENTIALS`: Path to a GCP service account JSON file (for Vertex AI). | ||
| - `VERTEX_PROJECT_ID`: (Optional) GCP project id for Vertex. If not provided, the SDK will attempt to infer it from credentials. | ||
| - `GEMINI_REGION`: (Optional) GCP region for Gemini/Vertex (defaults to `us-east5` or `CLOUD_ML_REGION`). | ||
| - `GEMINI_REGION`: (Optional) Vertex region override for Gemini calls (falls back to `CLOUD_ML_REGION`, then `us-east5`). |
There was a problem hiding this comment.
@caufieldjh @turbomam I did not notice this before, but is there a reason there are so many places to declare gcp region? can we change this to just GCP_REGION ?
There was a problem hiding this comment.
As with all things GCP, the answer is "yes and no". I think the values can be different (https://docs.cloud.google.com/gemini-enterprise-agent-platform/resources/locations kinda gets into this but doesn't fully answer the question).
For our purposes I think it's enough just to set the project ID and region.
There was a problem hiding this comment.
I guess I am asking because in this case they are filling the same value... to me this indicates they are not actually different. At least how we are using them
hesspnnl
left a comment
hesspnnl
left a comment
There was a problem hiding this comment.
This looks good - thanks for the initiative on this. I have one concern I commented above. I made some further docs changes that I had been wanting to do.
Closes #95.
Brings the README,
.env.example, and the code into agreement. The code reads env vars only throughos.environ; after this change, every variable it reads is documented and has a placeholder, and nothing it never reads appears in either file.cp .env-example .envto.env.example(the wrong name made the first quickstart step fail).DEFAULT_MODEL,MAX_TOKENS,TEMPERATUREfrom the README Configuration section (no code reads them).OPENALEX_API_KEYandUNPAYWALL_EMAILfrom.env.example(no code reads them, and there is no Unpaywall code despite the "required" comment). If either is planned, say so and I will restore it.CONTACT_EMAILandCLOUD_ML_REGION.NMDC_EDI_MAX_XML_CHARS,NMDC_DATAONE_SOLR_MAX_XML_CHARS,NMDC_HTTP_RETRY_ATTEMPTS,NMDC_HTTP_RETRY_BACKOFF_SECONDS,NMDC_HTTP_MAX_RETRY_DELAY_SECONDS,NMDC_HTTP_POOL_CONNECTIONS,NMDC_HTTP_POOL_MAXSIZE) to both, with their code defaults.tests/test_example.py.Co-Authored-By: Claude Opus 4.8 (1M context) noreply@anthropic.com