Upgrade to Pro — share decks privately, control downloads, hide ads and more …

CodeCity: On-Screen or in Virtual Reality? - VISSOFT 2021

CodeCity: On-Screen or in Virtual Reality? - VISSOFT 2021

35e56f49a18810a41205714698cd7e7d?s=128

David Moreno Lumbreras

October 06, 2021
Tweet

Transcript

  1. CodeCity: On-Screen or in Virtual Reality? David Moreno-Lumbreras - Escuela

    Internacional de Doctorado @ URJC - Bitergia Roberto Minelli - REVEAL @ Software Institute – USI, Lugano, Switzerland Andrea Villaverde - ETSIT @ Universidad Rey Juan Carlos Jesús M. González-Barahona - ETSIT @ Universidad Rey Juan Carlos Michele Lanza - REVEAL @ Software Institute – USI, Lugano, Switzerland VISSOFT’21
  2. Software World… CODECITY - Software systems as cities - Building

    features as metrics (height, area, color) - Building location as position of the artifact in the system hierarchy - Locality, orientation, habitability /city_metaphor
  3. Spiral algorithm - First element at the center of the

    spiral Interactive Building - Source code file (district as file hierarchy) - Base Area, Height and Color as metrics /babiaXR - CodeCity
  4. WebXR - Modern browser (including VR devices) OSS (A-Frame components)

    /babiaXR
  5. /experiment Research questions - RQ1: How does the accuracy of

    participants immersed in VR compare to that of participants using the on-screen version of BabiaXR-CodeCity? - RQ2: How does the efficiency of participants immersed in VR compare to that of participants using the on-screen version of BabiaXR-CodeCity?
  6. /experiment Target System - JetUML, two snapshots. - March 24,

    2021 - June 28, 2018 Metrics - Base area - Number of functions - Heigth - Lines of code per function - Color - Cyclomatic Complexity value (from blue to red scale)
  7. /babiaXR - CodeCity

  8. /experiment

  9. /experiment 24 participants - 12 on-screen (computer), 12 VR (Oculus

    quest 2) - Supervisor “with the eyes of the participant” - Academia and Industry
  10. Demographics survey 8 Tasks (the maintenance task definition framework by

    Sillito et al. [14].) Feedback survey /experiment
  11. /experiment

  12. Correctness - T1 and T5.1: 100% correctly answered /results

  13. Completion Time - Mann-Whitney U Test and Cliff’s Delta effect

    size /results
  14. Feedback - Location of the “core”: most of participants provided

    a reasonable answer. Some entered into more details highlighting specific directories. - Experiment not too difficult. - Few reported trouble moving in the 3D on-screen scene. /results
  15. Threats to validity (internal and external) Natural gestures in VR,

    not on-screen BabiaXR-CodeCity still prototype CodeCity in VR is comparable to CodeCity on-screen (multiplatform) /to conclude
  16. VR shorter completion time for the tasks, but less accuracy.

    In Top 5, always over 90%. Validate with larger samples Feedback = Improvement /to_conclude
  17. /replication package https://doi.org/10.5281/zenodo.5474890 https://thesis-dlumbrer.gitlab.io/icsme21-scenes/ ORO badge ROR badge https://babiaxr.gitlab.io/

  18. CodeCity: On-Screen or in Virtual Reality? David Moreno-Lumbreras - Escuela

    Internacional de Doctorado @ URJC - Bitergia Roberto Minelli - REVEAL @ Software Institute – USI, Lugano, Switzerland Andrea Villaverde - ETSIT @ Universidad Rey Juan Carlos Jesús M. González-Barahona - ETSIT @ Universidad Rey Juan Carlos Michele Lanza - REVEAL @ Software Institute – USI, Lugano, Switzerland VISSOFT’21