Post by @jonocarroll@fosstodon.org

2026-05-04 09:34 UTC

I&#39;ve been looking into dbt for data engineering and wanted to flesh out what the similarities and differences were to just using {targets} - I learned lots about both of them! I wrote up my findings in this post https://jcarroll.com.au/2026/05/04/comparing-r-s-targets-and-dbt-for-data-engineering/ #rstats #dbt #dataeng

Replies (3)

@defuneste@fosstodon.org 2026-05-04 16:01

@jonocarroll good post (and I 100% agree)! I have not try duckdb but my memory using PG is that with dbt you can also set up a staging / per user /branch. You are not just promoting code but also data (kind of). In fact you need it for validation. You can also do that with targets using multiple "stores". I think gittarget help here but I usually use my crappy function that switch store if not on main branch. I find validation tests slightly easier with dbt but never tried data.validator.

Open ##2044459
@tylermorganwall@fosstodon.org 2026-05-04 22:06

@jonocarroll you can separate out your components in {targets} as well if you create different files with sub-pipelines (lists of targets with assigned names), and source those files and add those lists to your main pipeline. I do that when my _targets.R file starts to get unwieldy.

Open ##2044460
@brianrepko@hachyderm.io 2026-05-05 16:51

@jonocarroll great post - I've also got on my todo list to look at mage.ai - https://www.mage.ai - since the coding blocks support #rstats and I got them to support renv::use() back in the day for packages needed by the block

Open ##2044479