windowrules no longer match correctly on current full window title

[MikeChristensen]

In previous versions of Hyprland, I was able to setup rules to remove window opacity on certain websites. For example, for YouTube:

windowrule = opacity 1.0 override 1.0 override,title:(.*YouTube*.)

In the recent versions of Hyprland, this rule no longer has any effect. I can, however, target the title of the process. For example this works:

windowrule = opacity 1.0 override 1.0 override,title:(.*Vivaldi.*)

However, now I'm applying that rule to all websites not just the ones I want such as YouTube.

According to hyprctl clients, here's the name of the window:

Window 564320c3b420 -> YouTube - Vivaldi:
        mapped: 1
        hidden: 0
        at: 1901,62
        size: 1515,434
        workspace: 1 (1)
        floating: 0
        pseudo: 0
        monitor: 0
        class: vivaldi-stable
        title: YouTube - Vivaldi
        initialClass: vivaldi-stable
        initialTitle: Start Page - Vivaldi
        pid: 3116
        xwayland: 0
        pinned: 0
        fullscreen: 0
        fullscreenClient: 0
        grouped: 0
        tags:
        swallowing: 0
        focusHistoryID: 2
        inhibitingIdle: 0
        xdgTag:
        xdgDescription:

This did work correctly when I was running NixOS 24.11 (not sure which version of Hyprland that was), but broke when I upgraded to 25.05 and remained broken in 25.11 which I believe is the latest stable Hyprland.

My Question: Did the logic for targeting windows by title change? Perhaps it uses the initialTitle rather than title now or something? It might make sense if this were some sort of perf optimization. Also, is there a way to target the current title of the window even if the process changes the title? Thanks!

Version Info

Hyprland 0.52.1 built from branch unknown at commit unknown unknown (unknown).
Date: unknown
Tag: unknown, commits: 0

Libraries:
Hyprgraphics: built against 0.4.0, system has unknown
Hyprutils: built against 0.10.2, system has unknown
Hyprcursor: built against 0.1.13, system has unknown
Hyprlang: built against 0.6.6, system has unknown
Aquamarine: built against 0.9.5, system has unknown

no flags were set

[vaxerski]

nothing should've changed but I rewrote rules in git so you might want to update, rewrite your rules for the new syntax, and recheck.

[Dregu]

Your regex is wrong and you're also reading the wrong version of the wiki.

[MikeChristensen]

Ah yes you're right about the Regex.. Good catch!

The following is working on 0.52.1:

windowrule = opacity 1.0 override 1.0 override,title:(.*YouTube.*)

I'm not sure why the docs say to use match:title, that does not appear to be valid syntax.

[MikeChristensen]

I still think something changed since version 0.49 because what I had used to work. Both (.*YouTube.*) and (.*YouTube*.) should match the window title of YouTube - Vivaldi according to this Regex tester which uses GoLang's Regex library (should be the same on Hyprland is using?).. The .* version would match the full string and the *. would match a single character after the e (including an end of line).

I'm wondering if a Regex title match needs to match the entire window title or if it can match part of the title. I'm also wondering if it used to work because I was using Brave before and maybe it generates window titles differently or something.

This might just be something I never figure out since I don't feel like going back to NixOS 24.11 and messing around with it. But the working Regex is arguably "more correct" since it'll match on any string that contains YouTube. It appears there's no bug here unless recent versions changed Regex matching rules or something..

[dcunited001]

I'm wondering if a Regex title match needs to match the entire window title or if it can match part of the title.

Yes, that was a fairly recent change.

i'm also a bit confused about the new match syntax.

• nixos machine with hyprland 0.52.0 and hyprlang 0.6.7 from flakes: really wants the new match: syntax
• and omarchy with hyprland 0.52.1 and 0.6.6: the match: syntax isn't recognized

I was wondering whether i can mix these hyprland/hyprlang versions ... but i think i just answered my own question

[MikeChristensen]

Yes, that was a fairly recent change.

Ok well now I know I'm not going insane. Mystery solved.

but i think i just answered my own question

Let me know if I'm following, but it sounds like the match:whatever changes are part of hyprlang 0.6.7, which would most likely show up in the latest stable NixOS channel some point in the future. However, the docs for latest Hyprland (0.52.1) mention this new syntax, even though most users wouldn't see this unless they're building hyprlang from source or using a custom flake.

That really makes the docs confusing because I'm like "Hey I'm using the latest Hyprland, cool I'll follow the docs here.."

[forbjok]

it sounds like the match:whatever changes are part of hyprlang 0.6.7

This is very strange.

I'm running hyprland v0.52.2 w/ hyprlang v0.6.7 on CachyOS, and it's not recognizing the "match:" syntax either.

[MikeChristensen]

Honestly I can't figure out what the whole match: thing is about.. There's only been a few changes to Hyprlang this year and none of them seem related..

[dcunited001]

@MikeChristensen it makes hyprlang more extensible, in the end, i think. i haven't used it yet (i downgraded for now)

• it used to be windowrule = rule, [,rules ...] [, params...]
• now they are windowrule = [ effect | match:params ]... with no order requirement (if i'm getting that right) and at least one param is required.

symbols from plugins (like names for effects/props) could collide with those from other plugins. probably a few other reasons maybe.

prefixing with a substring /always/ makes things more flexible though -- e.g. at least two DNS path components before TLD are required to support migration, certain types of redirection or SOA patterns. it's generally everywhere.

[dcunited001]

Also, named rules enable hyprctl keyword 'windowrule[my-rule-name]:enable false' which, on hyprctl reload can be reset to a known state. this makes some scripting much easier.

If i had to guess the reason, names can't be unambiguously defined without a multiline windowrule { name="fdsa" }. Either that feature or changing the existing windowrule syntax was needed. For various reasons, the underlying implementation for window rules would ultimately need to be changed. I can think of a few counter-arguments to that guess, but i'm pretty sure changes would require breaking the interface, so it's better that happens once, not twice

but ... idk

[dcunited001]

i would prefer having one simpler hyprlang version over supporting multiple concurrent versions. i think hyprland versions can run against multiple hyprlang versions though. i'm not sure how they interface together or how much version drift they'd tolerate.

i'd love to contribute to hyprland, but idk C/C++ and i'm not likely to learn overnight. I do want to do something with hyprwire though

[ailelix]

Guys I am using Hyprland 0.52.2-2 and Hyprlang 0.6.7-2 with Arch.
In the past I used to have rules for zen browser like:

windowrule=float, class:(zen) title:^(Picture-in-Picture)$
windowrule=pin, class:(zen) title:^(Picture-in-Picture)$
windowrule=size 480 270, class:(zen) title:^(Picture-in-Picture)$
windowrule=move onscreen 74% 74%, class:(zen) title:^(Picture-in-Picture)$

but now this cannot make the pic-in-pic window floating
I am also able to confirm that the match syntax is not allowed on my setup