Links to `p5.Element` don't work correctly

[marksherman]

Most appropriate sections of the p5.js website?

Reference

What is your operating system?

Mac OS

Web browser and version

Brave: 1.85.120 Chromium: 143.0.7499.192 (Official Build) (arm64)

Actual Behavior

Each page for a component that interacts with p5.Element, such as createButton, createSlider, createDiv, etc, and even createElement all have a bad link to the Element reference page, instead staying on current page.

This is located in the "Returns" section of the page, as each of these components return a p5.Element type object.

Each of these pages experiences the same issue of a malformed URL.

Create div has this URL: https://p5js.org/reference/p5/createDiv/
The link to the p5.Element is this: https://p5js.org/reference/p5/createDiv/#/p5.Element which does nothing but scroll elsewhere in the same page.

This is particularly frustrating while teaching in a classroom setting. The easiest way to get to the Element page is to search for "element" in the top bar and select "p5.Element" from the results which is unnecessary cognitive load and a huge interruption to flow of thought. The p5.Element page is necessary for all of these as that is where the method reference is location.

Expected Behavior

Under the "Returns" section of each of these pages the link to p5.Element should be https://p5js.org/reference/p5/p5.Element/

Steps to reproduce

1. Go to a reference page of a component that returns a p5.Element, which would be almost everything in the DOM section.
2. Scroll to "Returns"
3. Click the link in this line:

p5.Element: new p5.Element object.

Example (same as bove):

Page:https://p5js.org/reference/p5/createDiv/
The link to the p5.Element is this: https://p5js.org/reference/p5/createDiv/#/p5.Element which does nothing but scroll elsewhere in the same page.

Would you like to work on the issue?

sure but not really

[rakesh2OO5]

I can take this — looks like a reference link resolver bug. I’ll investigate where return types are being converted into anchors instead of reference routes.

[rakesh2OO5]

I’ve opened PR #1121 to fix the malformed p5.* links in the Returns and Parameter descriptions by normalizing #/p5.* anchors to /reference/p5/{type}/ at render time.

Verified locally on createDiv().

Would appreciate a review when you have a moment — thanks for the clear repro!

Before Fix :

After Fix :

[rakesh2OO5]

Thanks for the detailed breakdown!

I’ve applied link normalization specifically to @param and @return descriptions in ReferenceItemLayout.astro, which were the only sections not being rewritten before hitting the DOM.

Descriptions were already handled upstream, so the change is intentionally scoped to avoid altering routing behavior or introducing new paths.

Local behavior now matches production expectations (e.g. /reference/p5/p5.Element/ works, /reference/p5/p5.Element/center/ correctly 404s).

[ksen0]

Hi @marksherman @rakesh2OO5 ! Thanks for your patience. As I'm testing the PR I also see that createDiv() intro is formatted incorrectly (both in live site and the PR, so in any case it seems a separate problem). Can one of you let me know if this was like this when you first started working on the link fix, or if it's recent?

[rakesh2OO5]

Hi @ksen0 — thanks for checking.

This formatting issue with createDiv() existed before I started working on the link normalization fix.

I opened the PR on 29-01-2026, and I’ve attached a screenshot taken on that date (visible in the taskbar timestamp) showing the same formatting issue already present locally.

So this appears to be unrelated to the link normalization changes introduced in this PR.

Let me know if you'd like me to investigate this separately.

[ksen0]

Thanks so much for the data @rakesh2OO5 ! The source for the reference is in the main repository. The problem might be there, here, or anywhere along the way - I am not sure. I added more investigation notes for testing, and some contributor doc links. If you want to work on it, it's yours! If you find out that actually it's a website problem, please feel free to make the PR whereever is correct, and I can transfer the issue back if needed.