Appendix A — Appendix A: Setup

Account, roles, and first-run checklist

Keywords

snowflake, R, RStudio, Posit, VS Code, workspace notebooks, snowflakeR, RSnowflake, mlops

A.1 Snowflake account prerequisites

Before R packages can run, Snowflake roles need appropriate grants. Requirements vary by feature:

Feature Typical grants
SQL / RSnowflake USAGE on warehouse, database, schema; SELECT on tables
Workspace Notebooks Notebook execution, compute pool for ML Container Runtime
Feature Store / Registry ML object privileges per Snowflake ML RBAC
SPCS / doSnowflake USAGE on compute pool; CREATE SERVICE; image repository access
EAI bootstrap CREATE INTEGRATION or admin runs printed SQL

Work with your Snowflake admin to provision a development role with ML privileges before production promotion.

A.2 Local development setup

Full detail: Local R Setup, RStudio, Posit & VS Code.

pak::pak(c("Snowflake-Labs/RSnowflake", "Snowflake-Labs/snowflakeR"))

RSnowflake only — no Python step required.

snowflakeR — configure Python:

snowflakeR::sfr_install_python_deps()   # or pip/venv per chapter 03
snowflakeR::sfr_check_environment()

A.2.1 connections.toml

Create ~/.snowflake/connections.toml with at least one profile. Shared by CLI, Python, RSnowflake, and snowflakeR — see authentication options.

[default]
account = "ORG-ACCOUNT"
user = "MYUSER"
authenticator = "externalbrowser"
warehouse = "DEV_WH"
database = "DEV_DB"
schema = "PUBLIC"
role = "ANALYST"

Verify:

library(RSnowflake)
con <- DBI::dbConnect(Snowflake(), name = "default")
DBI::dbGetQuery(con, "SELECT CURRENT_USER()")

A.3 Workspace first run

Step Action
1 Create Workspace project; connect Git repo (optional)
2 Add snowflaker_config.yaml and reference sfnb_setup.py
3 Run Python bootstrap cell (setup_notebook)
4 Enable External Access in Snowsight if prompted
5 Run %%R smoke test (sfr_connect(), dbConnect(Snowflake()))
6 Open workspace_quickstart.ipynb as reference

Notebooks not next to sfnb_setup.py need /filesystem/ path discovery — Bootstrap — filesystem.

A.4 Optional: Posit Workbench Native App

If your account deploys Posit Workbench as a Native App, install the same R packages in the Workbench session and use connections.toml or inherited Snowflake auth — same APIs as desktop RStudio.