Special character in URL has new behaviour since 0.123.0

[jmichelgarcia]

What version of Hugo are you using (hugo version)?

latest (123.8)

When did the problem appear?

0.123.0

Problem

I am using Docsy and I have a file:

• content/en/docs/Administration/On-Premises Deployments/Architecture & System Requirements/_index.md

FontMatter wise:

---
linkTitle: "Architecture & System Requirements"
title: "Architecture & System Requirements"
weight: "10"
aliases:
  - /en/architecture---system-requirements.html
description: "xxx"
---

Until 0.122.0 that page was being rendered with this url:

• https://[domain]/en/docs/administration/on-premises-deployments/architecture-system-requirements/

After upgrading to 0.123.0, it is now being rendered as:

• https://[domain]/en/docs/administration/on-premises-deployments/architecture--system-requirements/

notice the -- instead of -.

Is this expected?

I'm going through the release notes and code, to see if I missed anything.
Many thanks!

[jmooring]

Reported here as well:
https://discourse.gohugo.io/t/hugo-v0-123-seems-to-anchorize-terms-in-their-urls-rather-than-urlize-them/48583

I know we made a "hyphen retention" change to address another issue (I think to retain three consecutive hyphens), but I'm not sure if the above is intentional.

[jmichelgarcia]

@bep, is the above here to stay (and therefore I should change my aliases, links and so on), or if this is an unexpected behaviour? Thanks.

[jmichelgarcia]

I believe this is the culprit (which I'm assuming is an intentional decision)

hugo/common/paths/path.go

Line 297 in 9ca1de0

wasHyphen = r == '-'

[jmooring]

Of a very small sampling (n=219) of sites in the wild running v0.100.0 or later, 23% are already running v0.123.0 or later, so... even if this wasn't intentional, changing it again may be unwise.

[jmichelgarcia]

If it wasn't intentional, I was hoping hugo would offer a way to opt-out (and preserve the old behaviour).

[jmichelgarcia]

@jmooring looking at it again, the current implementation doesn't make much sense though
it's basically rendering

• Architecture & System Requirements as architecture--system-requirements
• Installer (Kubernetes - OpenShift) as installer-kubernetes---openshift

Before I embark on a fixing journey, I wanted clarity on whether this is intentional or not.

[bep]

This is mentioned in the list of breaking changes (near the top) that was also in the release notes for v0.123.0:

#11455

#10104

[jmichelgarcia]

Thanks @bep and @jmooring

I think I'm an outlier because we have lots of pages with various special characters and spaces in them, which I will now correct.

I am however still puzzled on why the new behaviour is not consistent across different characters.
I did a quick check with Docsy and found the following:

filepath	url
content/en/docs/Concepts & Tutorials/_index.md	/docs/concepts--tutorials/
content/en/docs/Concepts - Tutorials/_index.md	/docs/concepts---tutorials/
content/en/docs/Concepts-Tutorials/_index.md	/docs/concepts-tutorials/
content/en/docs/Concepts (Tutorials)/_index.md	/docs/concepts-tutorials/
content/en/docs/Concepts (Tutorials/_index.md	/docs/concepts-tutorials/

Cheers.

[jmooring]

@jmichelgarcia We'll handle further questions here:
gohugoio/hugoDocs#2487

[github-actions]

This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.