From af9d32262c73d5f0bc98cbadbc31be0f15a6287e Mon Sep 17 00:00:00 2001
From: Marco Gorelli <33491632+MarcoGorelli@users.noreply.github.com>
Date: Thu, 3 Oct 2024 19:10:48 +0100
Subject: [PATCH] docs: add Ibis to "related projects"

---
 docs/roadmap_and_related.md | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)

diff --git a/docs/roadmap_and_related.md b/docs/roadmap_and_related.md
index 43580db6f..546279db8 100644
--- a/docs/roadmap_and_related.md
+++ b/docs/roadmap_and_related.md
@@ -29,3 +29,36 @@ Array counterpart to the DataFrame API, see [here](https://data-apis.org/array-a
 Allows C extension modules to safely share pointers to C data structures with Python code and other C modules, encapsulating the pointer with a name and optional destructor to manage resources and ensure safe access, see [here](https://arrow.apache.org/docs/format/CDataInterface/PyCapsuleInterface.html) for details. 
 
 Narwhals supports exporting a DataFrame via the Arrow PyCapsule Interface.
+
+### Ibis
+
+Pitched as "The portable Dataframe library", Ibis provides a Pythonic frontend
+to SQL. Some differences with Narwhals are:
+
+- Narwhals' main use case is for library maintainers wanting to support
+  different dataframe libraries without depending on any whilst keeping
+  things as lightweight as possible. Ibis is more targeted at end users
+  and aims to be thought of as a Dataframe library akin to
+  pandas / Polars / etc.
+- Narwhals allows you to write a "Dataframe X in, Dataframe X out" function.
+  Ibis allows materialising to pandas, Polars (eager), and PyArrow, but has
+  no way to get back to the input type exactly (e.g. there's no way to
+  start with a Polars LazyFrame and get back a Polars LazyFrame)
+- Narwhals respects input data types as much as possible, Ibis doesn't
+  support Categorical (nor does it distinguish between fixed-size-list and
+  list)
+- Narwhals separates between lazy and eager APIs, with the eager API
+  provide very fine control over dataframe operations (slicing rows and
+  columns, iterating over rows, getting values out of the dataframe as
+  Python scalars). Ibis is more focused on lazy execution
+- Ibis supports SQL engines (and can translate to SQL),
+  Narwhals is more focused traditional dataframes where row-order is defined
+  (although we are brainstorming a lazy-only level of support)
+- Narwhals is extremely lightweight and comes with zero required dependencies,
+  Ibis requires pandas and PyArrow for all backends
+- Narwhals supports Dask, whereas Ibis has deprecated support for it
+
+Although people often ask about the two tools, we consider them to be
+very different and not in competition. Further efforts to clarify the
+distinction are welcome 🙏!
+