From e7ee998f571afa9aabcc6fdf94ac26c19c891ec8 Mon Sep 17 00:00:00 2001
From: Harshal Sheth <hsheth2@gmail.com>
Date: Mon, 17 Jun 2024 13:50:08 -0700
Subject: [PATCH] fix(ingest): pin numpy<2 for classification (#10725)

---
 metadata-ingestion/setup.py | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/metadata-ingestion/setup.py b/metadata-ingestion/setup.py
index 7f51b39c2731b..04099e0a24b9f 100644
--- a/metadata-ingestion/setup.py
+++ b/metadata-ingestion/setup.py
@@ -107,6 +107,10 @@
     # This is a bit of a hack. Because we download the SpaCy model at runtime in the classify plugin,
     # we need pip to be available.
     "pip",
+    # We were seeing an error like this `numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject`
+    # with numpy 2.0. This likely indicates a mismatch between scikit-learn and numpy versions.
+    # https://stackoverflow.com/questions/40845304/runtimewarning-numpy-dtype-size-changed-may-indicate-binary-incompatibility
+    "numpy<2",
 }
 
 sql_common = (