Author Archives: Hannes

moOde audio player 8 on a Raspberry Pi 3 Model B Rev 1.2 with Hifiberry Amp2 1.1

It’s super easy but watch out because volume levels are messed up (super loud on tiny volume values!).

Flash the image with RPI imager (enable SSH)

/boot/config.txt might need dtparam=audio=off but I don’t remember, try without that change.

Boot the device

Visit http://moode.lan (or fix your network, use the IP etc) and click “m” -> “Settings”

“Audio” -> “Audio Output” -> “I2P”: Select “HifiBerry Amp2”


Use your favorite MPD client (or the web UI) to play some music (FIRST SET A LOW VOLUME for safety, 100 is insane here, ~20 is room level). It should work already, if not, troubleshoot.

“Audio” -> “ALSA Options” -> “Max volume (%)” seems to do nothing

“Audio” -> “MPD Options” -> “Volume options” -> “Max MPD volume”: Set to 30% so at least in the Web UI of moOde you won’t be able to kill your speakers.

In your other MPD clients, make sure you never set a high volume…

Music on connected storage will be discovered automatically and added to your library.

Remotely controlling QGIS through PostgreSQL queries 🤪

I just wanted to see if it was possible, it is not a sane thing to do.

PostgreSQL has a notification event system where you NOTIFY on a notification channel and clients can LISTEN for events.

QGIS already supports this for a convenient, remotely triggered refresh function of PostGIS layers by listening on the qgis channel if a PostgreSQL connection exists:

  • Load a PostGIS layer
  • Set its symbology to something dynamically randomized so you see changes if the map is refreshed
  • In the layer’s rendering options set it to refresh on notification
  • In a PostgreSQL client trigger a notification event, e.g. NOTIFY qgis, 'hi'; or SELECT pg_notify('qgis', 'refresh plz!');
  • The layer is refreshed in QGIS
  • You can set the layer(s) to refresh on any event on the qgis channel or only for specific messages. For example you could set each PostGIS layer to refresh on a different message, e.g. their table name and use that in a PostgreSQL trigger.

This is currently bound to existing PostGIS layers and as far as I know, you cannot do more than refresh layers. Edit: You can also trigger layer actions (thanks to the great people at Oslandia)!


QGIS comes with PyQt and PyQt (of course) has its own API to let you talk to database servers. And luckily its QSqlDriver class even has a notification signal that let’s you react to notification events in the most convenient way.

So I wrote something:

from qgis.PyQt.QtSql import QSqlDatabase, QSqlQuery, QSqlDriver

def exec_notification(name, source, payload):
    """ZOMG don't ever do this!"""
        exec(payload.replace('\\n', '\n'))
    except SyntaxError as e:
        iface.messageBar().pushMessage("PG NOTIFY", str(e))

db = QSqlDatabase.addDatabase("QPSQL")
db.setConnectOptions("service=foo")  # using ~/.pg_service.conf
if not
    raise Exception(db.lastError().databaseText())

    str, QSqlDriver.NotificationSource, "QVariant"

Run it in QGIS’ script editor and now you can remotely execute any Python code by sending it from PostgreSQL. For example:

SELECT pg_notify(
			'project = QgsProject.instance()',

SELECT pg_notify(

SELECT pg_notify(

Of course this is a horrible idea and you should never run code that a third-party sends to you. Instead you should write specific methods on the Python side and sent well defined messages to control exactly which methods get executed. But, hey it works!


  • Trigger a notification if a PostgreSQL-stored QGIS project gets updated and in QGIS notify the user if they have said project open.
  • Make a tiny plugin that lets the user specify a mapping of message -> Python code scripts
  • TBC


At first I had used a simple db.driver().notification.connect(on_notification), expecting the slot to simply receive all the arguments that the notification signal would send. But this led to Qt using a different notification signal (just sending the channel name) or something because the name is overloaded. Qt super explainer eyllanesc once again saved the day by showing that one can (and in this case needs to) specify the correct signature so Qt picks the correct signal to connect to. This was a major learning for me.

For debugging I used another slot function connected to the signal which uses both the QGIS message log and the message bar to inform the user about incoming notification events, take it if you want:

def on_notification(name, source, payload):
    message = f"{name} from {source}: {payload}"
    QgsMessageLog.logMessage(message, "PG NOTIFY")
    iface.messageBar().pushMessage("PG NOTIFY", message)

A bad surprise to me was that my QGIS installation on Ubuntu 22.04 (using did not have a dependency graph that included PostgreSQL support in QtSql. Ouch! I would have thought this would be usable in any standard QGIS installation out of the box. I had to install libqt5sql5-psql. On Windows this seems to be no issue, both a MSI installer and an advanced OSGeo4W install had the necessary modules. Phew!

Plugin options and QGIS3.ini

In the spirit of learning in the open:

All the docs say is “It is recommended to give some structure to naming of keys.”

The [PythonPlugins] section uses the plugins’ directory names and shows if a plugin is activated or not (true/false).

Plugins can store their options anywhere. No really. Anywhere! You will find some in their own sections, e. g.:

  • Qgis2threejs (directory “Qgis2threejs“) uses a [Qgis2threejs] section, named the same as its directory
  • First Aid (firstaid) uses a [FirstAid] section, using different case than its directory, and another section [firstaid] for other options
  • Plugin Builder 3 (pluginbuilder3) uses a [PluginBuilder] section, named differently than its directory
  • QGIS GML Application Schema Toolbox (gml_application_schema_toolbox) stores options in [QGIS%20GML%20Application%20Schema%20Toolbox] (needlessly) using URL encoding
  • Räumlicher Filter (spatial_filter) uses two sections: [SpatialFilter] and [SpatialFilterSymbol] (sorry, we didn’t know better…)
  • There is a [Plugins] and a [plugins] section. I found some plugins storing some options in the [plugins] section, e.g. First Aid storing some firstaid\debugger-geometry=... there. [Plugins] seems to be used by core plugins. [plugins] stores e.g. if QGIS should automatically check for new ones (although that is also might be tracked in the [plugin-manager] section, no idea which option comes from an older QGIS version maybe).

Some things can be stored elsewhere, I am not sure who does it, maybe some objects do this automatically? E.g. in the [QgsCollapsibleGroupBox] section you might find the states of all(?) QgsCollapsibleGroupBox objects such as gml_application_schema_toolbox\gmlas_bbox_group\collapsed=false concerning a plugin’s widget. Or in [Windows] you might find FirstAidDebugDialog\geometry=....

Plugins might alter any other options of course. There is no way of knowing what plugin touched what (great power and mostly good :) ).

Uninstalling a plugin does not remove the sections from the INI. The plugin itself does not have a chance to clean that up and QGIS has no way of knowing what to remove. (Of course it might be wanted to keep settings around, but it would be nice to be able to really prune something.)

tl;dr: There is no standard, no rules nor best practices for QGIS plugins to store their options.

PS: This post might get updated as I learn. Corrections and improvements are highly appreciated.

PPS: Excuse the big bold highlighting. That’s for me.

Backing up Hetzner snapshots locally

Hetzner is a nice, cheap host for server. Unfortunately they do not let you download backups and snapshots of cloud servers locally. So you are kind of locked in with them.

Here is how I create full disk backups for e.g. standard CX11 servers with default images.

Make sure you understand everything before you attempt this. Pay special attention to your own partitions and make sure you archive what you really need. Consider imaging the whole device instead of just a partition.

Create a backup

  • Create a new snapshot of your server
  • Create a new server from it
  • Boot said server in rescue mode
  • Use good old dd and gzip to image the main disk to a local archive
  • ssh root@SERVERIP "dd if=/dev/sda1 bs=1M status=progress | gzip -" | dd of=SERVERIP.sda1.dd.gz bs=1M
    • The bs=1M is fairly random by me, worked well, didn’t care to optimize
    • gzip (with the default compression level) kind of maxed out the server CPU but still enabled me to almost max out my download bandwidth. Of course the ratio between remote CPU compression speed vs download bandwidth depends on your specific situation. You might want to use xz or zstd instead.
  • Don’t be stupid, do it twice and compare the checksums

If you do not need your server to be live during backup, you can skip the snapshotting and second server of course, just boot your server in rescue mode instead.

Build a new server from a backup

  • Create a new server using the same or a similar configuration as the backed up one
  • Boot said server in rescue mode
  • cat SERVERIP.sda1.dd.gz | ssh root@SERVERIP "gunzip -c | dd of=/dev/sda1 bs=1M status=progress conv=fsync"
  • Reboot to leave rescue mode

If you want to restore your server because of a breakage, just boot it into rescue mode and do the same as above otherwise.

I can’t read file contents from my Android SD card on Linux and it’s weird

Conclusion: The files are encrypted with Android’s file-based encryption. The SD card is broken.

The SD card is broken. The internet says it is typical for SD cards to go read-only as a symptom of fail. Trying to rsync or ddrescue the card’s contents using a separate devices led to stalls and read errors.

I might be able to salvage some of the files by using Android’s share functionality but there was nothing important on the card so I won’t bother trying. The card is broken beyond repair and all files are lost.

I want to backup my Android SD card so I inserted it to my computer (tried multiple ones by now). I can see the file tree and browse it just fine. But if I try to read a file, it fails because instead of the actual file contents, all I get is some gibberish. Weirdly enough, the file header is partially identical between different files and contains the words “whatever” and “CONSOLE”. When inserted in my phone I can read and use those files just fine. The phone is a Sony Xperia XZ1 Compact.

Here are some example hexdumps of the file headers from some JPG and OPUS files:

00000000  00 00 00 00 00 37 d5 23  d8 c3 44 86 e4 42 f3 73  |.....7.#..D..B.s|
00000010  03 00 00 02 00 00 10 00  00 02 8c 2d 04 09 03 01  |...........-....|
00000020  77 68 61 74 65 76 65 72  60 ed 60 a5 16 dd d0 08  |whatever`.`.....|
00000030  34 0d 25 2b 87 d1 df 18  94 8a f8 cf f0 fd 83 d9  |4.%+............|
00000040  06 5a 4e 48 8c a1 b9 51  98 ed 16 62 08 5f 43 4f  |.ZNH...Q...b._CO|
00000050  4e 53 4f 4c 45 00 00 00  00 60 40 2d b1 41 51 c1  |NSOLE....`@-.AQ.|
00000060  10 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
00000070  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
00002000  c7 27 3f d6 7e 99 3d 6b  d3 dc 0a a7 c9 28 37 d5  |.'?.~.=k.....(7.|
00002010  59 09 8b 4b 3e 1b 20 54  a8 87 fc 90 fd 31 05 5b  |Y..K>. T.....1.[|
0037ffe0  f1 9a 2d fa fd a1 4b 2a  22 dc ce 29 9d 83 3a 5a  |..-...K*"..)..:Z|
0037fff0  dc 97 e0 e9 15 d7 16 55  82 a6 57 6b 7f b0 32 7d  |.......U..Wk..2}|
00000000  00 00 00 00 00 03 5d 1d  aa ab 3d 85 96 2a 8a 70  |......]...=..*.p|
00000010  03 00 00 02 00 00 10 00  00 02 8c 2d 04 09 03 01  |...........-....|
00000020  77 68 61 74 65 76 65 72  60 90 17 d4 8b ef 31 35  |whatever`.....15|
00000030  22 b9 b3 05 59 37 3b 71  7e e1 4a 6e af a2 07 b2  |"...Y7;q~.Jn....|
00000040  4b 9a bb 7e 6a 46 18 70  29 ed 16 62 08 5f 43 4f  |K..~jF.p)..b._CO|
00000050  4e 53 4f 4c 45 00 00 00  00 60 40 2d b1 41 51 c1  |NSOLE....`@-.AQ.|
00000060  10 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
00000070  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
00002000  32 23 5d 69 cb 14 d0 75  8a 9c 2e 64 67 26 1a fc  |2#]i...u...dg&..|
00002010  12 77 ad 48 9b f4 9e d4  d5 08 48 33 cb 45 6b 90  |.w.H......H3.Ek.|
00037fe0  87 09 0c 80 6c 4b 6a 54  fb c2 70 13 bc 0a c9 ab  |....lKjT..p.....|
00037ff0  89 3b 52 40 f2 cc df f4  d3 65 e1 c4 0d e3 74 ea  |.;R@.....e....t.|
00000000  00 00 00 00 00 01 78 41  33 26 6e 9c 0f a7 d9 69  |......xA3&n....i|
00000010  03 00 00 02 00 00 10 00  00 02 8c 2d 04 09 03 01  |...........-....|
00000020  77 68 61 74 65 76 65 72  60 2e 2d e8 1b cd 47 11  |whatever`.-...G.|
00000030  7e b6 e7 e2 95 84 85 75  81 42 0f 5c 54 48 3e 4b  |~......u.B.\TH>K|
00000040  c7 3c f6 cf 16 ec 9f 6b  51 ed 16 62 08 5f 43 4f  |.<.....kQ..b._CO|
00000050  4e 53 4f 4c 45 00 00 00  00 60 40 2d b1 41 51 c1  |NSOLE....`@-.AQ.|
00000060  10 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
00000070  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
00002000  97 9d 44 91 25 8e b4 e0  07 31 31 bc cc c0 44 13  |..D.%....11...D.|
00002010  62 0e 6d c1 49 8d 87 fe  5d c7 ff c0 cc b5 4c 08  |b.m.I...].....L.|
00019fe0  ae 2b d1 04 db d2 9c 9e  e7 cd 3a 06 aa 5a 85 5f  |.+........:..Z._|
00019ff0  d3 7a 30 33 f2 3a b1 2f  73 40 50 f7 a9 a6 d7 fa  |.z03.:./s@P.....|

fdisk shows the partition as “W95 FAT32 (LBA)“.

What’s going on?

I found some mentions of this issue on the web: posted “iu>i'Õè Œ- whatever`P‘Ã@C“Žâ=ë q¸ÌÔ9rÊ ¯Ø¶Þ¨Kíb_CONSOLE ]ŽÒ[Ó_” without a solution. posted “@Âù©õš•Bo   Œ-  `»Üž¸½\ht1Oc.ä9á²5Ñé’Ž<žŸ9Ûßíb_CONSOLE Øþ«fQ¿É“, also without a solution. The phone involved was a Samsung A3 2016. posted “#µ=Äõ¶EBC   Œ-  "3DUfw`â￾Lü4OOŸ-›·j.^aÌ?–ôaÚ ”YÜíb_CONSOLE .øŒ¤Èœ«é ‰'ñÏÆísÿä‰â¢THû´/Ò†´< bb·7iìÕ“, also without a solution. posted three examples, all also having at least the “b._CONSOLE” portion, Samsung Galaxy S3 device. No solution but a user commenting that “This is related to Samsung Knox and these signatures are Samsung specific” which seems unlikely as I am using a Sony device…

^ See the top of the post for my conclusion ^

Und tschüß, Forum/TPHH-Spiegel war ein (auf Basis von Discourse aufgesetztes) Forum für Code for Hamburg e. V., online vom Mai 2019 bis April 2023.

Die Homepage
Jeder Eintrag im Transparenzportal Hamburg wurde gespielt
Man konnte Suchen speichern und bei neuen Ergebnissen eine Mail erhalten
Ein Beispiel für einen Eintrag
Ein Eintrag mit mehreren verknüpften Resourcen

Die Idee war ursprünglich, dass dort Kommentare und Diskussionen zu den im Transparenzportal Hamburg (TPHH) veröffentlichten Datensätzen gepostet werden könnten. Und das vielleicht sogar direkt im TPHH eingebettet.

Dazu hat ein Python-Skript (als Bot namens “Heidi Kabel” ;) ) jede Minute die API des TPHH nach neuen Datensätzen gefragt, die Metadaten abgerufen und sie schön formatiert und aufbereitet als Foren-Threads gepostet. Ein spaßiges Bastel-Projekt, irgendwann 2019.

Die Dokumentation für die Forumsinstallation und der Code für den Bot liegen in:

Naja, dann kam Corona und Treffen von Code for Hamburg gab’s eh schon kaum noch. Die Seite hat praktisch niemand außer mir genutzt, es gab keine 10 Posts, die nicht vom Bot waren. Mit der Einbindung im TPHH wurde es auch nichts (ich hatte mich aber auch nicht drum gekümmert).

“Angesagte” Posts mit vielen Views
Pageviews exklusive (erkannte) Crawler
Top Referred Topics (keine Ahnung wo sie verlinkt wurden)
Der tägliche User war unser TPHH-Spiegel-Bot, die Posts/Topics auch

Ich hatte es für mich selbst noch als einfaches Interface für die Daten und als praktischen Benachrichtigungsservice auf bestimmte Keywords genutzt, aber jetzt klappt das Update der Forensoftware nicht mehr (Bug von Discourse mit angeblich nicht mehr einzigartigen Tags und es gab keine Rückmeldung auf meine Fragen, von daher …) und ich will da nix auf veralteter Software laufen lassen. Übernehmen wollte es niemand aus dem Verein und im Fediverse hat sich auch niemand gemeldet. Von daher mach ich die Tage das Licht aus.

PS: Ich hoffe Code for Hamburg wird irgendwann mal wieder reaktiviert, mit regelmäßigen Treffen, wo gehackt und gebastelt wird. Toi toi toi!

10 year old posts from an older blog at

Just for save-keeping, probably irrelevant nowadays but who knows what kind of travelers search engines might send here. Tread lightly!

Remove the huge margins from PDF papers

Remove the huge margins from PDF papers eg. prior to printing with pdfcrop from the pdftk suite. For example:

pdfcrop --margins -80 input.pdf output.pdf

edit, much better:

$ pdfimages -j file.pdf file
$ mogrify -crop 800x1300+220+220 +repage file-0*.jpg
$ convert file-0*.jpg cropped.pdf

+repage makes IM not write the offset to the files but actually crop “properly”. Maybe pdfcrop has such parameter too?

Installing a printer in Cups via a USB->Serial adapter

Make sure to “modprobe usbserial“. dmesg should show a printer being connected if you plug it in now. Then Cups should see it too.

RTTY with SDR# and fldigi (for the german DWD stations)

In SDR#: Use USB, filter bandwidth 1000. center the RTTY in your window.

In fldigi: Op Mode -> RTTY -> Custom. Set the carrier shift to custom and then enter 450 in the custom shift field below. Baud rate: 50, 5 bits per character, no parity, 1.5 stop bits. Save and Close. Make sure the Rv button is green!

CW decoding with SDR# and fldigi

In SDR#: Use the CW-L or CW-U preset. Tune so the morse code is right in the middle of your “reception window”.

In fldigi: Op Mode -> CW. Turn off squelch by making the SQL button not green but grey.

It works very well for non-human morse for me. Radio amateur morse is harder and so far full of “spelling errors”. :)

PDF to image with imagemagick/graphicsmagick

If you want to create images from PDF files, use for example mogrify -verbose -geometry 1600 -density 300 -format png *.pdf Without a decent “-density” parameter, you will probably get a blurry image as result.

Making your Ryzen CPU less hot by throttling boost on Linux

echo 0 | tee /sys/devices/system/cpu/cpufreq/boost

Might work, might not, depending on unknown factors. Whatever, I just wanted to make some all-core-but-unimportant process to run without going 95°C. For that it worked perfectly well. CPU temps of a Ryzen 5 3600 after many hours of full utilization were at ~65°C. CPU frequencies were capped to 3.6GHz with this while jumping up to 4.2GHz (and ~94°C) without.

Obviously this has an impact on performance.

To re-enable just echo a 1 instead. This is reset anyways when you reboot your system.

Properly setting up your QGIS license

If you want your copy of QGIS display it’s legal licensing status, this is the missing code for you.

Copy this in your QGIS Python script editor (WARNING: DO NOT RUN THIS IN AN IMPORTANT USER PROFILE, I will NOT help you if it breaks something):

import os
from qgis.core import QgsApplication, QgsSettings
from qgis.PyQt.QtGui import QColor, QImage, QPainter, QPen, QStaticText
from qgis.PyQt.QtWidgets import QMessageBox

def install_qgis_license():
    # WARNING: This fucks around with your profiles and stuff!
    # QgsCustomization is not available from Python so just yolo here and write a fresh file if possible
    profile_directory = QgsApplication.qgisSettingsDirPath()
    customization_ini_filepath = profile_directory + "QGIS/QGISCUSTOMIZATION3.ini"

    if os.path.isfile(customization_ini_filepath):
        # ain't gonna be touchin dat!
        text = (
            "Or: Custom license has been installed already...\n"
            "Anyways, we are not creating a *new* license now ;)"
        messagebox = QMessageBox()

    # get existing splash image
    splash_path = QgsApplication.splashPath()  # :/images/splash/
    splash_image_file = splash_path + "splash.png"
    splash_image = QImage(splash_image_file)

    # paint new splash image
    new_splash_image = QImage(splash_image)
    painter = QPainter()

    # white bold font plz
    font = QgsApplication.font()
    pen = painter.pen()

    # place text at appropriate location
    label_text = f"This QGIS©®™ is legally licensed to {os.getlogin()}"
    label_text_rect = painter.boundingRect(0, 0, 0, 0, 0, label_text)  # returns new rect that fits text
    left_offset = new_splash_image.width() - label_text_rect.size().width() - 20
    painter.drawText(left_offset, 840, label_text)


    # create license dir if necessary
    new_splash_dir_path = profile_directory + "license"
    except FileExistsError:

    save_success = + "/splash.png")
    if save_success:
        print(f"Initialized new QGIS license....")
        print("Error on QGIS license initialization, this will get reported!")

    # enable license dir for splash image lookup in QGISCUSTOMIZATION3.ini
    with open(customization_ini_filepath, "w") as sink:
        sink.write(f"splashpath={new_splash_dir_path}/")  # must have trailing slash

    # enable loading of QGISCUSTOMIZATION3.ini in user profile
    QgsSettings().setValue(r"UI/Customization/enabled", True)
    print("License installed, reboot QGIS to activate!")
    messagebox = QMessageBox()
    messagebox.setText("License installed, restart QGIS now to activate!");


Then (if you really want to do it), uncomment the function call in the last line and execute the script. Follow the instructions.

To clean up remove or restore the QGIS/QGISCUSTOMIZATION3.ini file in your profile and remove the license directory from your profile, restore the previous value of UI/Customization/enabled in your profile (just remove the line or disable Settings -> Interface Customization).

If you want to hate yourself in the future, put it in a file called in QStandardPaths.standardLocations(QStandardPaths.AppDataLocation) aka the directory which contains the profiles directory itself.

BTW: If you end up with QGIS crashing and lines like these in the error output:

Warning: QPaintDevice: Cannot destroy paint device that is being painted
QGIS died on signal 11

It is probably not a Qt issue that caused the crash. The QPaintDevice warning might just be Qt telling you about your painter being an issue during clean up of the actual crash (which might just be a wrong name or indentation somewhere in your code, cough).

Using a Tiardey USB Single Foot Pedal (PCsensor FootSwitch) on Linux

This post’s purpose is to link “Tiardey USB Single Foot Pedal Optical Switch Control One Key Programm Computer Tastatur Maus Game Action HID” to “PCsensor” and the footswitch tool on search engines so others who wonder if the device is easy to use on Linux learn that this is the case. Hope it helps!

I bought this which came with a chinese/manual saying “FS2007 User Manual” and also says “FS2007U1SW (mechanical switch)” (mine clicks, so I guess it is not the “FS2007U1IR (silent photoelectric switch)”. The manual links to for Windows drivers.

Plugin the device. dmesg should show something like:

[Sun Jan 8 20:25:05 2023] usb 1-4: new full-speed USB device number 7 using xhci_hcd
[Sun Jan 8 20:25:05 2023] usb 1-4: New USB device found, idVendor=1a86, idProduct=e026, bcdDevice= 0.00
[Sun Jan 8 20:25:05 2023] usb 1-4: New USB device strings: Mfr=1, Product=2, SerialNumber=0
[Sun Jan 8 20:25:05 2023] usb 1-4: Product: FootSwitch
[Sun Jan 8 20:25:05 2023] usb 1-4: Manufacturer: PCsensor
[Sun Jan 8 20:25:06 2023] input: PCsensor FootSwitch Keyboard as /devices/pci0000:00/0000:00:14.0/usb1/1-4/1-4:1.0/0003:1A86:E026.0001/input/input19
[Sun Jan 8 20:25:06 2023] input: PCsensor FootSwitch Mouse as /devices/pci0000:00/0000:00:14.0/usb1/1-4/1-4:1.0/0003:1A86:E026.0001/input/input20
[Sun Jan 8 20:25:06 2023] input: PCsensor FootSwitch as /devices/pci0000:00/0000:00:14.0/usb1/1-4/1-4:1.0/0003:1A86:E026.0001/input/input21
[Sun Jan 8 20:25:06 2023] hid-generic 0003:1A86:E026.0001: input,hidraw0: USB HID v1.11 Keyboard [PCsensor FootSwitch] on usb-0000:00:14.0-4/input0
[Sun Jan 8 20:25:06 2023] input: PCsensor FootSwitch as /devices/pci0000:00/0000:00:14.0/usb1/1-4/1-4:1.1/0003:1A86:E026.0002/input/input22
[Sun Jan 8 20:25:06 2023] hid-generic 0003:1A86:E026.0002: input,hidraw1: USB HID v1.10 Device [PCsensor FootSwitch] on usb-0000:00:14.0-4/input1
[Sun Jan 8 20:25:06 2023] usbcore: registered new interface driver usbhid
[Sun Jan 8 20:25:06 2023] usbhid: USB HID core driver
[Sun Jan 8 20:25:10 2023] usb 1-4: reset full-speed USB device number 7 using xhci_hcd

Sweet, so it is just some rebranded PCsensor device.

lsusb says ID 1a86:e026 QinHeng Electronics FootSwitch btw.

There is a great little tool for configuring those on Linux:

footswitch -m ctrl -k 1 will configure it to send Ctrl+1 when pressed for example. See the readme for usage and more examples.

You can use more than 3 of these devices via this pull request. I have four connected via a USB hub 1a40:0101 (“Terminus Technology Inc. Hub (branded “hama”), and they work just fine.