Section: New Results

Teaching of Algorithms, Programming, and Debugging

Participants : Florent Bouchez-Tichadou, Theo Barollet, Aurelien Flori, Thomas Herve.

Teaching Algorithms using Problem and Challenge Based Learning

Teaching algorithms is always a challenge at any level of the CS curriculum, as it is often viewed as a theoretical field. While many exercises revolve around classical examples that illustrate interesting algorithmic points, they are often disconnected from reality, which is a major drawback for students trying to learn. During the last four years, we have been trying to reconnect the teaching of algorithms with their applicability in the real world to M1 and L2 students, by giving them actual problems that could arise in their life of future software engineers, challenging enough to force them to use particular algorithmic techniques or data structures—e.g., linked lists, binary trees, dynamic programming or approximation algorithms.

By assigning students in groups of 5 to 6 members, we wanted to create an environment where they function as a team trying to work together to solve a problem. This allowed them to help each other in their respective comprehension, and made them more autonomous in their learning. The effective materials was provided as online pdf files so they had to read and learn from them by themselves, while the class sessions with a tutor (teacher) where used for the problem-solving part, with guidance from the tutor (who is there to make sure the learning takes place).

After four years of experimentation with M1 students, we found that the student's grades were stable, in particular there was no decrease in exams performances compared to the classical course that was taught in the previous years. However, the students progressed in trans-disciplinary skills such a communication and the writing of essays. More importantly, students show a strong adhesion to the teaching method, 50% of them rating it as “excellent” (6) and 25% as “good” (resp. 6 and 5 on a scale from 1 (terrible) to 6 (excellent)). No student rated the course below average.

This work has been published in the 23rd International Conference on Innovation and Technology in Computer Science Education, ITiCSE 2018.

Data Structures Visualization at Runtime

Debuggers are powerful tools to observe a program behaviour and find bugs but they are not often used by developers and especially beginners because of the hard learning curve of such tools. They provide information on low level data but are not able to analyze higher level elements such as data structures. This work tries to provide a more intuitive representation of the program execution to ease debugging and algorithms understanding.

We have a basic prototype, Moly, which is a GDB extension (GNU Project Debugger) to explore a program runtime memory and analyze its data structures. It also provides an interface with an external visualizer, Lotos, through a formatted output. Running Moly along with a dedicated visualizer should allow a programmer to spot bugs easier by seeing the subsequent whole memory states of the program and some data structures information.

The current status of Moly allows a programmer to explore all attainable memory at any point during the debug process, and already provides minimum information about the possible properties of the data structures, such as recognizing graphs, trees, or linked lists. Future work includes recognizing access patterns to the structures to extract for instance visit patterns and higher-level properties (such as the breaking of data structure properties between break-points).

The external visualizer, Lotos, is still in its early stages of development and was enough to make a proof-of-concept that it is possible to display via a web browser the information gathered by Moly. Our plans is to redesign this part from scratch using the knowledge gaining during the writing of this prototype.

AppoLab: an Online Platform to Engage Students in Their Learning

Classical teaching of algorithms and low-level data structures at the L2 european level is often tedious and unappealing to students, with much of the time being spent on analysing and devising algorithms for textbook cases, such as sorting lists of integers, visiting linked lists or trees, etc.

Using Problem-Based Learning helps to alleviate this problem, by presenting more complex problems to handle, hence engaging more students in their learning. This work revolves around the design of a learning platform that includes gamification in PBL. AppoLab is in its core a server that has scripted "exercices". Students can communicate with the server either manually, using telnet; but ultimately, they will need to script the communication also from their side, since the server will gradually impose constraints on the problems such as timeouts or large input sizes.

This preliminary work was used this year in some parts of an Algorithm course at the L2 level, and has received positive feedback from the students. This encourages us to continue this development and study more precisely the impact it has on students' engagement in their learning.