Dmitriy Kostiuk

An explanation of the structure of any complex system becomes more easily digestible and less confusing if it is accompanied by simple and understandable diagrams. Different subject areas have their own ways of depicting objects. In the case of Kubernetes clusters, this task becomes very non-trivial. Traditional network diagrams are a poor fit for a container orchestrator; official emblems have features that require their careful and moderate use. Finally, the complexity of Kubernetes makes it very easy to overload a drawing with details. We will look at how Kubernetes clusters are usually drawn by various authors. It will show how auxiliary tools such as color coding, grouping, and eye anchoring can make the cluster diagram more understandable... and how not to get the opposite effect when drawing your cluster.

Estimating the user's mental state with a set of special measuring devices can be helpful in detecting bottlenecks of the human-computer interaction. Until recent years, electroencephalography devices were too expensive and too complicated for most UX researchers, but now there are affordable consumer-grade EEG devices.

The talk covers EEG headsets produced by NeuroSky and Emotiv, as well as the open hardware OpenBCI project. Each headset has its advantages and disadvantages for UI/UX research. Specifics of data that can be acquired from each headset is reviewed, and existing open-source tools and libraries to get these data are discussed. The talk explains how we can use EEG headsets to evaluate mind concentration and relaxation, as well as rejection and arousal factors, which can be used to detect positive and negative emotions. EEG usage scenarios are discussed with examples of the FLOSS projects exposed to such UI testing.

Estimating the user's mental state with a set of special measuring devices can be helpful in detecting bottlenecks of the human-computer interaction. Until recent years, electroencephalography devices were too expensive and too complicated for most UX researchers, but now there are affordable consumer-grade EEG devices.

The talk covers EEG headsets produced by NeuroSky and Emotiv, as well as the open hardware OpenBCI project. Each headset has its advantages and disadvantages for UI/UX research. Commercial devices have different primary goals complicating their usage for research with cyphering, special licenses and limitations in the vendor-provided proprietary SDKs - but open-source tools developed by the community improve the situation. OpenBCI is quite the opposite: it is fully open, but much harder to obtain/build.

Specifics of data that can be acquired from each headset is reviewed, and existing open-source tools and libraries to get these data are discussed. The talk explains how we can use each headset to get information about the user's mental state. Mind concentration or relaxation is a highly informative parameter for UX; it can be easily evaluated and it can be obtained with any of the EEG headsets. Besides that, most of the reviewed devices provide enough data to measure rejection and arousal factors, which can be used to detect positive and negative emotions. Finally, EEG usage scenarios with examples of the FLOSS projects exposed to such UI testing are discussed.

Running old-time OS in a virtual machine often comes with a problem of non-coinciding cursors. A relative positioning device (mouse and trackball) provides the operating system with a vector of the cursor movement instead of its coordinates. As a result, different cursor acceleration formulas are breaking the coincidence of host and guest systems cursor. Fortunately, modern guest operating systems support absolute positioning devices (tablets), which provide real cursor coordinates. Desktop virtualization systems emulate such devices (USB Wacom tablet in case of QEMU), allowing the host cursor to control guest systems in so-called "mouse integration mode".

Until recent times mouse integration was available only if guest OS either has a driver from the virtual machine vendor or supports USB tablet. In all other situations, desktop virtual machines use "mouse lock mode". The host cursor is hidden, and the user can only interact with the guest system until some dedicated keypress.

Starting from version 1.9, QEMU had accepted our GSOC-developed patch intended for the osimeline project (HTML-based live demo of historical operating systems) and can emulate the RS-232 Wacom tablet as a "wctablet" character device backend. Currently, QEMU is the only virtualization system with such capabilities.

Guest OS still needs such tablet drivers, which exist at least for DOS, OS/2, Windows 3.x and Windows 95, BeOS, Mac OS 9.x, Mac OS X, and pre-USB Linux (some drivers are even FLOSS). Drivers for Windows 3.x Windows 95 and Linux are the most compatible ones and are easiest to configure. The talk will show how to activate mouse integration with the legacy OS. The situation with wctablet-compatible drivers and the possibility to improve it will be discussed as well.

Estimating the user’s physical and mental state with a set of special measuring devices can be helpful in detecting bottlenecks of the human-computer interaction. Until recent years, evaluating cognitive and physical load by biometrical parameters (heart rate, galvanic skin response, brain waves, gaze direction, etc.) was too expensive to be widely adopted for FLOSS. However contemporary consumer-grade gadgets targeted at fitness and entertainment are much more affordable and precise enough to be used in the UX/UI comparison. Still, their different primary goal often complicates their usage for the research. The talk will highlight which devices are the most suitable ones for the research purposes in the open-source world (the ones having open-source and GNU/Linux frameworks to access biometric data). Gadgets covered with the talk are fitness-trackers, EEG headsets, and eye-trackers. Patterns of getting data, problems with cyphering and licensing will be discussed, as well as brief biometry usage scenarios and examples of the UI express-testing.