LoginSign Up
← Back to all sparks
D

dbplyr

ANALYTICS
Velocity0.0

Database backend for dplyr that translates R code into SQL queries

github.com/tidyverse/dbplyr ↗

dbplyr ends its two-year backend migration by dropping 1st edition support outright

sqldplyrdatabase-backendsbreaking-changestidyverse
◆Current state
dbplyr translates dplyr code into SQL, and 2.6.0 closes a migration that has been running since 2023: first-edition backends no longer work at all. The same release converts a long list of soft deprecations into hard failures and removes functions deprecated as far back as 2019. The releases before it were translation-quality work across SQL Server, Redshift, Snowflake, Postgres, Spark and Teradata.
◆Where it's heading
The package is trading compatibility surface for a smaller, more consistent core it can actually evolve — qualified table names were overhauled in 2.5.0, sql() and ident() were refactored internally, and the cte argument gave way to a single sql_options() entry point. Backend breadth keeps growing at the translation level even as the extension API narrows.
◆Prediction
With the edition split finally gone, expect the next cycle to spend its budget on dialect translations and the newer Spark/Databricks path rather than on further deprecation.

◆Recent moves

  1. 1mo ago

    1st edition backends removed; a wave of deprecations goes defunct

    ⚡ SPARK

    The release the last two years were building toward. Every remaining first-edition backend stops working, previously-warning deprecations become errors, and long-dead functions are deleted — a deliberate reduction of the surface dbplyr has to keep translating.

    View source ↗
  2. 11mo ago

    Date and aggregate translation fixes across six SQL dialects

    Corrects difftime() sign handling and adds date_count_between() on SQL Server, Redshift, Snowflake, Postgres and Spark, and moves SQL Server to COUNT_BIG so count() survives large tables. Argument checks are tightened enough that some previously silent failures now error.

    View source ↗
  3. 2y ago

    Qualified table names overhauled; I() becomes the simple path

    Rewrites how schemas and catalogs are specified, restoring ident_q() and in_catalog() usage and removing the unsuppressable warnings that came with the previous attempt. Also adds clock function translations across four dialects.

    View source ↗
  4. 2y ago

    Preliminary Databricks Spark SQL backend; join fixes

    Adds the Spark SQL backend that later releases keep refining, alongside a batch of breaking argument renames in the sql_query_*() family and repairs to full_join, semi_join and Oracle connections.

    View source ↗
  5. 2y ago

    dbplyr 2.3.4

    A hot patch for CRAN check failures with no functional change.

    View source ↗
  6. 3y ago

    dbplyr 2.3.3

    A hot patch for CRAN check failures with no functional change.

    View source ↗