[fix] Cache layout_config data and use during replay #12183

sfc-gh-lwilby · 2025-08-12T15:32:30Z

Describe your changes

While working on width for st.image I discovered that since we only cache the inner element proto (e.g. Image.proto, Code.proto), when an element with width/height added in the new style on Element.proto is cached it does not replay with the correct width/height.

This PR caches the layout_config as well so that it can be replayed as well.

GitHub Issue Link (if applicable)

Testing Plan

Unit Tests (JS and/or Python) ✅
E2E Tests ✅

Contribution License Agreement

By submitting this pull request you agree that all contributions to this project are made under the Apache 2.0 license.

github-actions · 2025-08-12T15:32:45Z

✅ PR preview is ready!

Name	Link
📦 Wheel file	https://core-previews.s3-us-west-2.amazonaws.com/pr-12183/streamlit-1.48.0-py3-none-any.whl
🕹️ Preview app	pr-12183.streamlit.app (☁️ Deploy here if not accessible)

snyk-io · 2025-08-12T15:42:20Z

🎉 Snyk checks have passed. No issues have been found so far.

✅ security/snyk check is complete. No issues have been found. (View Details)

✅ license/snyk check is complete. No issues have been found. (View Details)

sfc-gh-lwilby · 2025-08-12T18:00:23Z

lib/tests/streamlit/runtime/caching/cache_data_api_test.py

+
+        @st.cache_data
+        def cache_code_with_layout():
+            st.code(


I debated covering more elements (especially since this is how this gap slipped by originally), but I think it would be excessive. I chose st.code because it has height and width.

sfc-gh-lwilby · 2025-08-12T18:01:35Z

e2e_playwright/st_cache_data_test.py

    expect(image_element).to_have_attribute("src", image_src or "")
+
+
+def test_cached_code_replay(app: Page, assert_snapshot: ImageCompareFunction):


Adding this with st.code because 1) It has both height and width 2) It doesn't have width/height on it's inner proto like st.video and st.audio.

Copilot

Pull Request Overview

This PR fixes a caching issue where layout configurations (width/height) for Streamlit elements were not being preserved during cache replay. Previously, only the inner element proto was cached, causing cached elements to lose their layout styling when replayed.

Key changes:

Enhanced the caching system to store and replay layout configuration data alongside element protobuf data
Updated cache replay functionality to apply layout configurations during element reconstruction
Added comprehensive test coverage for both @st.cache_data and @st.cache_resource decorators

Reviewed Changes

Copilot reviewed 7 out of 13 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
`cached_message_replay.py`	Added layout_config field to ElementMsgData and updated replay logic to apply layout configurations
`__init__.py`	Updated save_element_message function signature to accept layout_config parameter
`delta_generator.py`	Modified _enqueue method to pass layout_config to caching system
`cache_data_api_test.py`	Added unit tests for layout config preservation during cache replay for @st.cache_data
`cache_resource_api_test.py`	Added unit tests for layout config preservation during cache replay for @st.cache_resource
`st_cache_data.py`	Added cached code element with width/height for e2e testing
`st_cache_data_test.py`	Added e2e test verifying visual consistency of cached elements with layout configs

Copilot · 2025-08-12T18:32:58Z

lib/tests/streamlit/runtime/caching/cache_resource_api_test.py

+        actual_height = element.height_config.pixel_height
+        expected_msg = f"Expected {description.lower()} height {expected_height}, got {actual_height}"
+        assert actual_height == expected_height, expected_msg
+


The _assert_layout_config helper method is duplicated between cache_data_api_test.py and cache_resource_api_test.py with identical implementation. Consider extracting this to a shared test utility module to avoid code duplication.

Suggested change

Might do this. I didn't already because we don't already have a test_utils file for this area and I wasn't sure it completely necessary for just these two tests.

Yeah, I think we could refactor it to shared parameterized replay tests. But also fine to leave it as is.

lukasmasuch · 2025-08-13T08:53:58Z

e2e_playwright/st_cache_data_test.py

+    code_element = app.get_by_test_id("stCode").first
+    expect(code_element).to_be_visible()
+
+    # Test dimensions with snapshots since the width/height is set on the element container.
+    assert_snapshot(code_element, name="st_cache_data-st_code_before_caching")
+
+    click_checkbox(app, "Show code")
+    expect(code_element).not_to_be_attached()
+
+    click_checkbox(app, "Show code")
+    expect(code_element).to_be_visible()
+    assert_snapshot(code_element, name="st_cache_data-st_code_after_caching")


nit: since this test is just about ensuring that width & height are correct, I think we can just do a to_have_css("width", expected_width) here instead of snapshot testing (which is more expansive and easier to miss when updating lots of snapshots).

The stCode element doesn't have the CSS styles, it's on the Element Container. Maybe adding some test utility to get the associated element container is a good idea to reduce some snapshot testing.

Another option would be to get the bounding box, e.g.:

streamlit/e2e_playwright/shared/vega_utils.py

Lines 18 to 26 in 0d8dfa4

def assert_vega_chart_height(vega_chart: Locator, expected_height: int):

vega_graphics_doc = vega_chart.locator("[role='graphics-document']")

bbox = vega_graphics_doc.bounding_box()

assert bbox is not None

assert round(bbox["height"]) == expected_height

def assert_vega_chart_width(vega_chart: Locator, expected_width: int):

lukasmasuch

LGTM 👍

persist layout config during caching and use in replay

cc3ef1a

sfc-gh-lwilby added security-assessment-completed Security assessment has been completed for PR change:bugfix PR contains bug fix implementation impact:users PR changes affect end users labels Aug 12, 2025

update snapshots

afea48d

sfc-gh-lwilby marked this pull request as ready for review August 12, 2025 17:34

sfc-gh-lwilby commented Aug 12, 2025

View reviewed changes

sfc-gh-lwilby requested a review from Copilot August 12, 2025 18:32

Copilot AI reviewed Aug 12, 2025

View reviewed changes

lukasmasuch reviewed Aug 13, 2025

View reviewed changes

lukasmasuch approved these changes Aug 13, 2025

View reviewed changes

sfc-gh-lwilby merged commit 506bd8b into develop Aug 13, 2025
43 of 45 checks passed

sfc-gh-lwilby deleted the fix/layout-config-cache-replay branch August 13, 2025 12:17

		expect(image_element).to_have_attribute("src", image_src or "")


		def test_cached_code_replay(app: Page, assert_snapshot: ImageCompareFunction):

	def assert_vega_chart_height(vega_chart: Locator, expected_height: int):
	vega_graphics_doc = vega_chart.locator("[role='graphics-document']")
	bbox = vega_graphics_doc.bounding_box()

	assert bbox is not None
	assert round(bbox["height"]) == expected_height


	def assert_vega_chart_width(vega_chart: Locator, expected_width: int):

[fix] Cache layout_config data and use during replay #12183

[fix] Cache layout_config data and use during replay #12183

Uh oh!

Conversation

sfc-gh-lwilby commented Aug 12, 2025

Describe your changes

GitHub Issue Link (if applicable)

Testing Plan

Uh oh!

github-actions bot commented Aug 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

✅ PR preview is ready!

Uh oh!

snyk-io bot commented Aug 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🎉 Snyk checks have passed. No issues have been found so far.

Uh oh!

sfc-gh-lwilby Aug 12, 2025

Choose a reason for hiding this comment

Uh oh!

sfc-gh-lwilby Aug 12, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Copilot AI Aug 12, 2025

Choose a reason for hiding this comment

Uh oh!

sfc-gh-lwilby Aug 12, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lukasmasuch Aug 13, 2025

Choose a reason for hiding this comment

Uh oh!

lukasmasuch Aug 13, 2025

Choose a reason for hiding this comment

Uh oh!

sfc-gh-lwilby Aug 13, 2025

Choose a reason for hiding this comment

Uh oh!

lukasmasuch Aug 13, 2025

Choose a reason for hiding this comment

Uh oh!

lukasmasuch left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

github-actions bot commented Aug 12, 2025 •

edited

Loading

snyk-io bot commented Aug 12, 2025 •

edited

Loading

sfc-gh-lwilby Aug 12, 2025 •

edited

Loading