The issue seems to be Keysharp interfering!

Original post:

I have this set up as a startup script, but it only sometimes triggers, despite setting a 20-second delay:

#!/bin/bash
sleep 2
xmodmap -e "clear lock"
xmodmap -e "keycode 66 = Return"

I have the startup command set to bash "path/Remap_Caps_Lock_to_Enter.sh". What's going on? It works when I press the triangle button when it doesn't work on boot.

The first and foremost I see unclear is the PATH and PWD. That is, the environment. For starters:

1. Have you tried passing full paths to the executables?
2. Have you tried recovering the PATH with /usr/bin/env via Shebang?

What if the "start" script doesn't find the locations due to missing definitions/locations as a result of different environment if compared to your shell/process you test on?

Boot logs?

Does it run but error or not run at all?

I'm a newb! How do I check boot logs? However, I'm not sure of how relevant those may be (also @Artwork@lemmy.world - I don't know if that tagged correctly, but I hope so); I just discovered another mysterious piece of the puzzle: it initially works on login, but about 5 min in, something actually undoes the mapping and I have no idea of what; rerunning the script fixes it until the next shutdown.

Other remapping/expansion tools I use are Keysharp and Espanso, if that helps to know, but I don't have them interacting with Caps Lock... I would like to more easily manage this just through Keysharp, but Keysharp currently can't seem to block the native function of Caps Lock from simultaneously occurring.

Hmm, another thought I had is that I wish startup times would have no maximum delay; Mint Cinnamon's max is a 100-second delay; why?... That's annoying. That alone could possibly solve this (even if just as a workaround).

Yes, tagging is valid, and I got a notification. Thank you!

I am sorry, but it seems like we have yet to know the actual operating system (OS), distribution, actual paths for the scripts, and Init system (e.g., Systemd, Initd etc.). Did you use Cron, or modify shell configs? We do not know your unique local customized environment, and changes you've done, yet, which is required for more precise investigation.

I have almost no idea of what half of what you're saying is; I've heard of systemd, initd, and cron, but I don't know what exactly they are and I don't know what "shell configs" are.

I thought it was obvious that I'm running Linux Mint since I posted this in !linuxmint@programming.dev. I'm using Zena, 22.3 on kernel 7.0.0-14, which seems to be the latest one that Update Manager is showing me.

1. Create the file /home/user/another_folder/remap_caps_to_enter.sh containing the code above
2. Open Startup Applications and set it to bash the above path as a command
3. It initially works, but then after a few min, it reverts to native Caps Lock behavior. Manually rerunning the file again makes it work for the rest of the session until I shut down the PC.

I see. Thank you for the details shared!

I've never experienced Linux Mint, but a quick Google shows an option "Caps Lock acts as Return" in the desktop environment (DE) settings. Is that the main idea?

Regarding the custom script case. Is it possible a system-wide service executes after your boot/startup script, slightly later, remapping it again?

We may try investigating. For instance, in a new terminal on the very system start, via xev, to try capture the moment it changes:

xev;

Or manually:

#! /usr/bin/env bash

declare s='' s2='';

while :;
do
    s="$( xmodmap -pke | grep -E 'keycode\s+66'; )";
    
    if [[ "$s" != "$s2" ]];
    then
        printf -- ' [%s] Changed from "%s" to "%s".\n' "$( date -- '+%F_%H-%M-%S'; )" "$s2" "$s";

        s2="$s";
    fi

    sleep 1;
done

I would also try checking out logs:

journalctl --user -f | grep -E 'settings|keyboard|layout|xkb';

# Or: journalctl -f | grep ...

-- xev -event mapping;
++ xev;

a quick Google shows an option "Caps Lock acts as Return" in the desktop environment (DE) settings.

That's a figment. It doesn't exist. I had checked the list of possible reassignments and was furious that Enter is, like, literally the only reassignment that isn't in the list. I mean, unless something recently got added, it wasn't there a month ago. That'd be sweet if it is now...

I'm not at a PC right now but will check later, thanks.

If I am not mistaken, Linux Mint settings service uses xkeyboard-config.

Would you mind to add "caps:return" to the list of supported keyboard options?

Source: https://gitlab.freedesktop.org/xkeyboard-config/xkeyboard-config/-/work_items/121 [2017-04-07]

---

xkeyboard-config 2.43...

Added caps:return to make the Caps Lock key an additional Return key.

Source [2024-10-01]

---

<option>
  <configItem>
    <name>caps:return</name>
    <description>Make Caps Lock an additional Return key</description>
  </configItem>
</option>

Source

The actual commit: 6b30f36201b40fddcf28918eb4c96b13a23b42ae

Yet, I've checked in a virtual machine online, and there's no such, apparently, indeed (sorry).
However, I believe it's possible to modify it manually in file /usr/share/X11/xkb/rules/evdev.xml, and I've tried quickly the following (after changes to the file, restart the csd-keyboard; e.g., by killing it). Would it work in your case (I would also try rebooting after)?:

Oh, sweet, huh, nice. Thanks a ton for the research; I totally didn't think to go this route. I'll try it when I next can!

Update: So, @Artwork@lemmy.world (I don't know if this will notify you), unfortunately, I did get around to finally trying it but even though the XKB option is picked, it doesn't seem to actually do anything despite rebooting... but I sure appreciate you poking around and trying; I didn't even know that file existed or could be modified.

xev: unrecognized event mask 'mapping'

I forgot to check these before already applying it... It's okay... I'll just tolerate it for now. I may try the other code later but am wiped out... Thanks for all the ideas, though.

Sincere apologies! Apparently, the mapping was in my dated local custom build of the tool...
Have you tried recording the general xev (with no , without any explicit filtering?

Your xev build/Man/Info message should list the supported. For example, in v1.2.3, these are supported:

$ man xev;
# ...
-event event_mask  
When not specified, all events are selected... Available event masks: keyboard mouse expose visibility structure substructure focus property colormap owner_grab_button randr button

Yet, today we found the actual possible culprit, and monitoring the events is probably not required. We know that it will just change, and the issue source should be a separate whole system-wide service that changes your initial change to its own configured, eventually.

There are many custom key remapping tools, including keyd, but I would try realizing/experiment with the standards first, and I appreciate you for still trying! It may help you in the future.

Wonderful day! Yes, thank you! I got the notification for the edit! Marvelous Lemmy!

The main idea is that it feels like a keyboard layout system service executes after your initial script and hence you notice no difference.

I am sorry, but I am not sure I'll be able to support the case at this moment, or at least a week, thoughtfully enough to be supportive, considering the job and time availability... I am sorry...

Meanwhile, have you considered responding in the forum above that restarting the services, after editing the file, and select the option, do not affect the key events processing?

Also, have you tried clarifying with other experienced people at IRC? Some networks, as Libera (most people from Freenode) and OFTC, may have those who have more time right now, and possibly had your very case, too?:
- https://libera.chat/guides/clients#terminal-clients
- https://www.oftc.net/
- https://netsplit.de/channels/?chat=mint

I've been known on Libera with a nickname as Lore (e.g., channel #linux), but they banned me again, long ago, unfortunately, for trying to protect myself against trolls... so please be careful! Yet, there are indeed truly ingenious, incredibly people!
Oh! Please also consider sharing the solution when found!

Dang, sorry to read about your ban. By the way, I think it is Keysharp that is causing the problem because it seems to be working just fine when I have that app disabled. I do like it a lot though, so I'll keep tinkering and finding some way to possibly try to get both of them to work concurrently. It uses Microsoft's .NET 10 runtime, but the latest versions have been having issues on my PC, so I've been going without Keysharp, and the Caps Lock reconfiguration is working flawlessly, haha. Thanks for your patience!

Thank you for the update! Just to clarify, I am sorry, what is KeySharp exactly? Is it one of the following?:
- https://github.com/goaaats/KeySharp
- https://github.com/mfeemster/keysharp
- https://apps.microsoft.com/detail/9nkmq5kbjg59

None of the above, ironically; it's: https://github.com/Descolada/Keysharp

Thank you very much for sharing!

AutoHotKey (AHK) are quite great for Windows, yet I am not sure about Linux, where you may have even more featureful and optimal solutions based on Xorg/Wayland, and/or your desktop environment (e.g., KDE Plasma).
Unless you don't want/have time to migrate these Windows-oriented scripts, are you sure you want to depend on them?

I ported over as much as I could to Espanso, yeah. I appreciate your concern and would ideally ditch this software too but the problem is that I haven't found a really competent equivalent; Python is just way too tricky to figure out and libraries like pywinauto or whatever it's called are really limited in terms of what they're capable of doing (not enough!), or AutoKey is just insanely complex with so many files being needed versus just one master file that one can use with AutoHotkey (which I also use at my Windows-mandated workplace).

For example, one frequently used script of mine in AutoHotkey/Keysharp purges the clipboard's URL of all known trackers before pasting, even doing HTML-decoding as needed if encoding is detected. I haven't been able to find anything that can as easily do this on Linux, though I suppose I could poke around more, or if worse comes to worst, try to have an AI build a cross-platform one, haha.

As we know, it depends on the actual environment, where simpler or more complex options are available out-of-the-box.
Meanwhile, have you tried?:
- https://github.com/hluk/copyq
- https://copyq.readthedocs.io/en/latest/scripting.html

In CopyQ, since its compiled with the clipboard event handling, it's sure possible to handle these events, including changing the content or launching custom scripts, and though I don't like Python, I do like JavaScript, and CopyQ has built-in support for quite limited but JavaScript with its own API function built-in to handle it, mentioned in the documentation above.

Preview of the settings I've just checked out in a Kubuntu machine I had on hands:

And it worked! The quick scripts I tried are the following:

copyq:
try {
    var original = str(clipboard());
} catch(e) {
    abort();
}

// If empty or too long
if (!original || original.length > 10000) {
     abort();
}

try {
    var executed = execute('/home/user/scripts/clean-url.sh', original);
} catch(e) {
    abort();
}

// If failed to execute
if (!executed) {
    abort();
}

var cleaned = str(executed.stdout).trim();

// If empty or matches the original (CopyQ may get into a loop otherwise)
if (!cleaned || !cleaned.length || cleaned === original) {
    abort();
}

// Change the clipboard buffer
copy(cleaned);

That program or script, to clean it, may be anything you prefer of course, including Bash, Perl, PHP, JavaScript, or what AutoHotKey does/executes in your case possibly.

There's the following to automate the read/write I know but haven't used, too:
- https://github.com/Slackadays/Clipboard

In Wayland, you may use wl-paste to read, and wl-copy to write into the clipboard.
In X11, xclip to both read and write.

The thing is that there should be some kind of event to handle on the clipboard change, and there options as DBus or Qt, yet I am not currently aware of a verified option to do so without an actually compiler program to subscribe to such an event.

For instance, Klipper, the Qt program, uses the even handling over DBus exposed via its libraries for convenience.