To be able to compare our findings with existing pressure curves of a previous approach [1], logging our data correctly is inevitable. Due to not influencing our test results we implemented two threads for logging pressure and event data. These log files contain id, timestamp and pressure / event data. Furthermore, we no longer write our data to a file during the logging process. As this approach takes too much time, we switched to storing our output to a struct variable. This struct is written to a file, when the measurement is done. To distinguish the different button presses, we integrated an empty line between two press processes. To check wether the actual events correspond to the amount of button presses, we implemented a third thread for event counting. After implementing these threads, we successfully tested the logging and calibration with a Keyboard. The log files were correctly created and filled with the determined data.

After consultation with our supervisor, we discarded our first idea (see last entry) and decided to focus on two new approaches. On the one hand a vacuum based fixture and on the other hand two clamps holding the input device in place. The vacuum based holder will most likely be discarded in consequence of the following two aspects: We found no good technical instruction to build such a construction and even if this happens, we probably will not have enough time for that. The second approach consists of two clamps [2] attached to the tabletop and holding the device in place. We already have found the needed clamps and its extensions. Therefore, we only need to buy and assemble these to a working construction. However, right now we prioritize this task lower than the actual measurement, so a solution in the near future is not to be expected. .

The Graphical User Interface processed further. We implemented log4j and a logging to each view. Speaking of views, two new views were added. First is the bigger log, which will help with the readability of logs. This is due to the bigger screen and more items. The other is the chart view, which shows a linechart with each data point of a .data file. This data file contains pressure data and will be read to find out the time a pressure point was pressed. So this view will visualize the curve to get a better understanding of the data and check whether its a curve or not.

Now we are beginning to test the validity of our implementation. Is it compliant to the previous tests with an invasive approach? We will see. Furthermore, we are going to check whether the java based gui has any influence on the latency. Additionally, we will investigate, if different button types will have any influence on the pressure curve. An automated mode for comparing the non-invasive data with the invasive data is also considered.