ICSE 2021 Co-located events: Mining Software Repository(MSR)

Day1

Posted by Bhagya C on May 17, 2021

Today I attended first day of ICSE 2021 co-located events. As my research is ML for Software Engineering I decided to have a keen attention of MSR events. It is 3 day event I will keep you guys posted on some fancy, and important research happening in the industry. As the elders say, we have to keep up with the trend otherwise we will get outdated. So here we are!!! 😎

Identifying vefrsions of Librares Used in Stack overflow code Snippets

This paper was authored by Ahmed Zerouali (Vrije Universiteit Brussel, Belgium), Camilo Velázquez-Rodríguez (Vrije Universiteit Brussel, Belgium), Coen De Roover (Vrije Universiteit Brussel, Belgium) a I have the access to the entire paper proceedings as I have registered to the conference 😅🤣. But do not worry I will address the main topics. And if you want to know more let me know we can learn together.

 Idea 💡

The purpose of the paper is self explanatory. We all rely on stack overflow most of the time we will get the solution as code snippets which is added by some authors. But as a user who accessing it later in the time might not know the version. And it can cause issues in our integration. This project is a solution to this. It will predict the version of the code snippet.

Implementation 🛠

In this paper they are exploring the version of Maven libraries referenced in Java code snippets on Stack overflow

They collected all the libraries available, extracted the JAR files from MAVEN for each version and extracted class and method names.

Also they extracted the java code libraries that make use of Maven libraries in the Stack over flow code snippets from stack overflow torrent

Then searched for the used library classes and methods in the JAR file and found the possibly used library version

For each used library they found compactible range of compactilbe  with classes and methods used in the code. Each range has two ends, most recent(best casse) and last (worst case).

And the main take away is that even after the last update the Stack Overflow answers comd with code snippets making use of outcated 

My first thoughts on this paper 🧐 ( PS:I have not read it in depth so don’t judge me)

Pros:

  • Very important insight for the developers before blintly taking codes from the codebase and understanding the recent versions

Cons:

  • But very sensitive to the given library one have to recreate entire process for each and every languages and classes available. 🤯