[Enhancement] Allow searches, for letters with diacritics, to not match letters without.

[RokeJulianLockhart]

When I searched squarefree.com/#:~:text=%C3%A2, I expected it to match the content of squarefree.com/#post-844:

I have mixed feelings about requiring Mozillians to “agree� to the Mozilla Manifesto. I get the impression that many volunteers aren’t fond of “commercial involvement� (9).

Instead, it matches the first mere “a” within the <H1> of the website.

My Environment

1.4.xpi, on firefox-nightly-152.0a1-20260429203445.

[ichaoX]

This is a feature and complies with wicg.github.io/scroll-to-text-fragment

[RokeJulianLockhart]

@ichaoX, apologies. Can the user somehow except their text fragment from this?

I ask because guidance that is phrased as vaguely as “this is a case-insensitive search also ignoring accents, umlauts, and other marks” is not desirable, for me.

[ichaoX]

I’m not against adding an option to handle diacritics, since ignoring them definitely makes things more complex and less reliable. However, non-standardized features could cause inconsistencies across different environments.

To improve matching precision, try using text fragments with context. For just a quick search, Ctrl+F works fine.

Out of curiosity, were you using #:~:text= as a search tool?

[RokeJulianLockhart]

I’m not against adding an option to handle diacritics, since ignoring them definitely makes things more complex and less reliable. However, non-standardized features could cause inconsistencies across different environments.

@ichaoX, I've filed WICG/scroll-to-text-fragment/issues/268. I imagine that it'll be rejected, but worth an attempt, perhaps?

To improve matching precision, try using text fragments with context. For just a quick search, Ctrl+F works fine.

Out of curiosity, were you using #:~:text= as a search tool?

I might have been, depending upon how one interprets my usage of it. However, I would err otherwise: I was attempting to demonstrate, to someone else, the commonality of those erroneous, diacrticised characters, across a webpage, whose source markup wasn't publicly available.

A text fragment was the best way to link to these occurrences, and because so many existed, explicitly generating a directive that includes a range for every occurrence would make less sense than merely encoding the character in the fragment.