Collection of resources: corpus linguistics, e-learning, natural language processing, teaching English as a foreign language and tools related to these topics
In combination with Tailscale (and if self-hosting is preferred/important Headscale) it offers secure and easy GUI access to computer and virtualised desktops in a homelab (plus cloud) setup:
No server is needed for direct peer-to-peer connections (e.g. with tailscale full domain names or with internal IPs in homelab):
NB: Don’t forget to enable 2-Factor-Authentication.
* on high-resolution/Retina displays: use this workaround ** currently on TestFlight only
If you’d like to enable external access to resources in your homelab or in your cloud environment, please don’t forget to choose a reasonably secure setup:
For month I’ve been trying to get this to work and I finally found a solution, which is a lot easier than I initially thought it would be.
Problem: If you want to connect remotely to a Linux desktop (running ThinLinc server) via ThinLinc* client, it has often been impossible to get the remote desktop working at the native resolution of a 4K display (3840×2160). In my case the highest possible resolution to choose from in the remote systems’ display settings was 1920×1080, resulting in a blurred linux desktop. While ThinLinc still offered a superior remote desktop experience to all tools tested, the blurred display kept bugging me. My aim was to get the sharpest and most responsive Cendio ThinLinc sessions on high-resolution displays possible.
NB. The following workaround also helped me solve display resolution problems with other remote desktop clients (e.g. NoMachine, VNC variants).
[Update 19/02/2025]: The best alternative to ThinLinc for connections to ARM-based linux and macOS remote desktops (e.g. RasperyPI, Oracle Free Tier Ampère Instances, Apple’s M1/M2/M3/M4) is RustDesk, which is available for all platforms and allows peer-to-peer connections (ip<->ip) without the need for a server if used in small homelab situations. Unfortunately, thinlinc server is only available for Intel-/AMD-based systems.
RustDesk does offer superior handling of display resolutions, however, for remote desktops on Intel-/AMD-based platforms thinlinc is by far the best solution for connections to Linux desktops. In these cases, I prefer to use the following workaround (also see overview of my favourite RDP client<->server combinations)
Solution on Windows 11 and macOS with 4K external display and Mac retina display
(might also work on other high-resolution displays)
Step 1 – Set display scaling of the display(s) you want to use for your remote connection to 100% prior to initiating the remote connection
On Windows:
Right-click on Windows-Desktop and choose: “Display settings”. Scroll down to scaling and make sure that your system does not use “custom scaling”. Choose 100% from the drop-down menu. Alternatively, you can navigate to Windows Settings > System > Display settings to find the same menu.
(Screenshots are from a German version of Windows 11)
NB: If you are using more than one display, make sure to adjust scaling to 100% on all screens you intend to use for your remote connection. Also make sure that there is no “custom scaling” enabled.
On macOS:
On macOS, display scaling can be set under: System Preferences/Displays:
Choose the setting labelled “More Space” for native retina display’s resolution.
NB: If you are using more than one display, make sure to adjust scaling to 100% on all screens you intend to use for your remote connection.
Step 2 – Now, start thinlinc client and connect to your remote linux desktop:
(Don’t be discouraged by the tiny icons and font size, you can switch back to your normal scaling as soon as your thinlinc session is over.)
Step 3 – On the remote system, you now have the opportunity to choose your native display resolution and adjust scaling to a more convenient factor:
Respondus 4.0® is a powerful tool for creating and managing exams that can be printed to paper or published directly to Canvas, Blackboard, Brightspace, Moodle, and other learning systems. Exams can be created offline using a familiar Windows environment, or moved between different learning systems. Whether you are a veteran of online testing or relatively new to it, Respondus 4 will save you hours on each project.
Extremely promising new Python NLP tool: spaCy (commercial open-source software):
Unfortunately, it is only able to deal with English input at the moment and installation on Windows seems to be tricky. The project is currently under intense development and it will be interesting to check the following links on a regular basis:
