Redirecting to homepage broken for version switcher

[lagru]

The documentation for the version switcher states

When clicked, the switcher will check for the existence of that page, and if it doesn’t exist, redirect to the homepage of that docs version instead.

I think I discovered at least two cases where this doesn't seem to work:

• On https://scikit-image.org/docs/0.21.x/release_notes/index.html switching to any prior version gives a 404.
• On https://pydata-sphinx-theme.readthedocs.io/en/stable/examples/index.html switching to v0.9 gives a 404.

We might have some wrong configuration but finding this issue on your website as well leads me to believe that this may actually be a bug?

[lagru]

cc @stefanv

[12rambau]

is it related to #932 ?

[12rambau]

note that for our theme it's expected as there were no "example" section in v0.9 (see: https://pydata-sphinx-theme.readthedocs.io/en/v0.9.0/index.html)

[lagru]

I don't think this is a duplicate of #932, though it might be related. #932 describes behavior that happens on the root of v0.8.0, while this issue pertains to the switchers behavior on subpages.

note that for our theme it's expected as there were no "example" section in v0.9

Precisely. The fact that v0.9 doesn't have an example section is why we can see the supposed bug in action. The cited part from the doc pertains to the behavior of the version switcher if the switched-to version doesn't have a corresponding equivalent of the current subpage.

[12rambau]

the code dealing with this is store here:

pydata-sphinx-theme/src/pydata_sphinx_theme/assets/scripts/pydata-sphinx-theme.js

Line 289 in dbab287

function checkPageExistsAndRedirect(event) {

I tried to reproduce the configuration in of v0.9 in a codepen (https://codepen.io/12rambau/pen/RwqwdvV) and it should return the correct homepage.

My best guess is that the location is not updated and the event returns the original url whatever the reult of the fetch test. As I'm no js expert I need to continue investigate.

[12rambau]

related to #844

[lagru]

Hey, thanks for trying to addressing this. I just tried to test the fix with the approach described below, and it seems like it still doesn't work?

1. Setup scikit-image (assumes a clean Python environment outside the project directory)

git clone https://github.com/scikit-image/scikit-image.git
cd scikit-image
pip install -r requirements.txt -r requirements/build.txt

2. Apply the following patch

Subject: [PATCH] Update pydata-sphinx-theme to rev e8d1e68
---
Index: pyproject.toml
IDEA additional info:
Subsystem: com.intellij.openapi.diff.impl.patch.CharsetEP
<+>UTF-8
===================================================================
diff --git a/pyproject.toml b/pyproject.toml
--- a/pyproject.toml	(revision a20870ee582f844482096e2edff65fa9e511213c)
+++ b/pyproject.toml	(date 1687369797918)
@@ -101,7 +101,7 @@
     'kaleido',
     'scikit-learn>=0.24.0',
     'sphinx_design>=0.3',
-    'pydata-sphinx-theme>=0.13',
+    'pydata-sphinx-theme @ https://github.com/pydata/pydata-sphinx-theme/archive/e8d1e689105dd135762a021821f8348845c38172.zip',
 ]
 optional = [
     'SimpleITK',
Index: requirements/docs.txt
IDEA additional info:
Subsystem: com.intellij.openapi.diff.impl.patch.CharsetEP
<+>UTF-8
===================================================================
diff --git a/requirements/docs.txt b/requirements/docs.txt
--- a/requirements/docs.txt	(revision a20870ee582f844482096e2edff65fa9e511213c)
+++ b/requirements/docs.txt	(date 1687369796044)
@@ -17,4 +17,4 @@
 kaleido
 scikit-learn>=0.24.0
 sphinx_design>=0.3
-pydata-sphinx-theme>=0.13
+pydata-sphinx-theme @ https://github.com/pydata/pydata-sphinx-theme/archive/e8d1e689105dd135762a021821f8348845c38172.zip

3. Build documentation

spin docs

4. Open the resulting file doc/build/html/index.html inside a browser and navigate to the "About" section in navigation bar, try to switch to version 0.20. Which attempts to access the non-existent https://scikit-image.org/docs/0.20.x/about/index.html.

I am not sure whether there it makes a difference that I am accessing the site on the local file system instead of some server.

[drammock]

darn, yeah, I was afraid that #1350 would not be enough.

[12rambau]

Note that I'm also super sade ;-)
but TBH I really don't understand why it's not working. if I check the header of the 404 page it should work. Looking more closely to the inspector of my browser, it seems the link is followed before the try/catch statment, maybe the preventdefault method should be called at the begining of the function and not at the end ?

(sorry I'm no javascripts expert, it's try and error process here)

[12rambau]

Removing the block-release impact as it's not preventing core functionality and we need to release the latest layout. I'm still trying to understand why links are followed, but I'm afraid I need more JS skills for that.