mranderson17

the qobuz webapp is hi-res too, I just use it in Firefox and my dac reports the same bit/sample rate that qobuz does. AFAIK there's no compression there though I haven't extensively verified that, only that the end result is 24bit/192kHz if that's what qobuz says is playing.

EDIT: Also, qobuz is nice because there's very few things you can click on in the web interface which cause the music to stop playing. I really appreciate that feature.... looking at you bandcamp....

Ah, I should have been more clear. The CRITERIA section of the sway documentation states that class matches support regex, so instead of using a * as you did in your example you'd use a regex any .*. So I think (untested of course) that for_window [class="steam_app_.*"] allow_tearing yes should work.

The comment in the code for allow_tearing notes that it must be enabled on the output as well. Here is the relevant output documentation. There are several other notes/recommendations there as well you should probably pay attention to.

Are your games all wine/proton games? For me in sway they all have the same class followed by some uid thing:

] > swaymsg -t get_tree
[...]
  #92: output "DP-5"
    #70: workspace "21"
      #126: con "Automobilista 2" (xwayland, pid: 171976, instance: "steam_app_1066890", class: "steam_app_1066890", X11 window: 0x5400001)

Or gamescope:

] > swaymsg -t get_tree
[...]
  #92: output "DP-5"
    #70: workspace "21"
      #124: con "Assetto Corsa" (xdg_shell, pid: 170694, app_id: "gamescope")

EDIT: Also allow_tearing was added to master 3 weeks ago, so this is definitely not in the current release. FYI to anyone who might try it.

Or add EA anti-cheat 6 months after release like they did for EA WRC.

Yeah, it changes without skipping a beat for me in pipewire, even in things like zoom/teams.

I use a little oneliner with tofi (rofi/wofi would also work) to select the current output and avoid pavucontrol. It's mapped to a sway binding but would probably work in any wm/de:

pactl set-default-sink $(pactl list short sinks |awk '{print $2}' |tofi $tofi_args)

I'm using pipewire so the functionality of pactl is actually provided through pipewire-pulse I think

Gatgetbridge (your link) has a breakdown of devices they support https://gadgetbridge.org/gadgets/ . You can click through the vendors to find devices which are both "highly supported" and "no vendor-pair". Meaning most/all the features work without any reliance on the vendor app.

As for the similarity you are asking about with pixel->GrapheneOS, there are very few watches that can run an alternative open source firmware or operating systems apart from the ones that are already open source, like bangle.js, pinetime, etc. Wearables are even more specialized than phones, they require specialized code designed specifically for them and would likely require pretty extreme effort to reverse-engineer.

I use a pebble 2 HR with gadgetbridge but the watch it self runs the old pebble firmware which gadgetbridge talks to. This is fine for me, but if you are looking for a more modern watch you may have to make some compromises.

The SQLite database is encrypted, though there was a period of time where it wasn't I think which may persist if your DB is older, but the key is stored right next to it on the filesystem. Signal desktop doesn't use your keyring or any of the other available methods to unlock it's database which is why you don't have to enter anything when starting the application, and why you can move it between machines by simply copying the .config/Signal dir. So while they are "encrypted", it's effectively clear text if you have access to the directory the database is stored in.

What makes those methods better?

Disclaimer, I'm at a sortof "advanced hobbyist" level of cad. My understanding of the topological naming problem in general is that it exists in all cad because it is a sortof byproduct of how computers keep track of data about 3 dimensional objects. If you make a cube, all the sides need to have an identifier associated with them. If you put a hole in that cube, you now have more identifiers and have to decide what ordering makes sense. It sounds easy to work around with a cube but when models get really complex it's not so easy, especially when you change something way back at the beginning which creates more or less faces in the middle of the list somewhere.

Freecad isn't making the topological naming problem "go away". They are creating (or rather merging, it's been around a long time) an algorithm that makes a better guess at what the order should be, rather than sticking new faces in the list and reordering without any consideration of what happened after that face was created. This is, as far as I understand, also how other CAD packages do it, and you can still back yourself into a topological naming problem if you try hard enough (or don't try at all I guess) in both freecad with the new changes applied, and in other CAD packages.

So "best practice" is to be smart about the attachment of your geometry thinking about how things might change in the future, rather than clicking the closest face whenever you need a sketch plane. In reality modern proprietary cad is so good at guessing and maintaining consistency that it doesn't matter unless your model is horrendously complex and whoever made it didn't pay any attention to laying out the base sketches in an organized way.

For example if you make a flange but you're not quite sure about the thickness, base the sketch for say, the holes, on the parallel origin and offset it by the height of the pad or the length of the sketched geometry. Or use a spreadsheet or variableset for the value of both the thing that you define the thickness with, and the offset from the origin plane. That way if the value changes, nothing will break.

I made a test model but it isn't something that shows up well in a single screenshot unfortunately. See the "Flange Thickness" and z offset parameters in the property view. I used that for the flange dimensions, and the hole sketch offset.

Interesting, it is working for me in wayland and the drop down menus are fine but I'm using sway which is a totally different wayland implementation than what KDE is doing. I'm glad you found a workaround.

Unfortunately I don't know what is causing the exact issue you are having, however here are a few things I found when doing this myself that are "gotchas" (not immediately obvious).

1. This is the reason your fonts are all Times New Roman. Go to that key using protontricks regedit and delete all the font replacements.

2. Anything you put in $HOME/.steam/root/steamapps/common/assettocorsa stays there, even if you uninstall the game. If you want to "start over" you have to uninstall the game and then delete the whole assettocorsa directory there, and the wine prefix in $HOME/.steam/root/steamapps/compatdata/244210

3. AC and content manager work without .net changes in the latest GE but you do need corefonts which you can install with protontricks. If you want to be extra sure you have the right .net you can install dotnet472 but I don't believe this is necessary anymore as it will be installed automatically or is already installed. You may get a wine .net error the first time you launch the game but it's only the first time.

4. If you choose to use CSP you have to unzip the archive you get from either Patreon or acstuff.ru and manually copy the dwrite.dll file into $HOME/.steam/root/steamapps/common/assettocorsa on EVERY upgrade. The zip installer built into CM doesn't do this correctly on Linux. It will cause rain not to work if you choose to use the Patreon version if you don't do this manual step.

I think you should start over and make sure the assettocorsa directory is clean before re-installing the game. It could be missing fonts, but it's hard to say. You can back it up somewhere if you have data in there you need.

I like tofi as a wayland alternative. Unfortunately like so many of these projects it seems to be somewhat recently unmaintained.