Versioning

Keith VanderLinden
Calvin University

Project Management

We’ll be adapting the use of project management tools developed for software engineering.

• Artifacts: Git/GitHub
• Tasks: GitHub Projects
• Time: Clockify

• Overview: CS and DS projects are:
  • Different in that they focus on different things and manage different artifacts (q.v. module 1).
  • Similar in that they both must do project management (q.v. modules 2–3).
• We will re-purpose SE project management tools for use in ML/Data Engineering.
  • DS artifacts are still drafted, edited, maintained over time, delivered.
  • DS projects must still manage multiple, self-interacting tasks done by multiple, collaborating people over time.
• Schedule:
  • DVCS - this week - managing artifacts over time
  • Task/Time management - next week - managing people over time

Version Control Systems

• Version control systems (VCSs) are used by teams to manage artifacts over time.
  • Centralized VCSs maintain a single, central repository of shared artifacts.
  • Distributed VCSs maintain a copy of the repository on each client machine.

• Version control systems (VCSs) are used by teams to manage artifacts over time.
  • Teams - multiple people working together on a project.
  • Artifacts - code; datasets; notebooks; documentation; images; presentations; etc.
  • Time - a version history of artifacts is maintained over time
• DVCSs are (most?) common, but some companies are using CVCSs, e.g.:
  • Azure DevOps Server (https://docs.microsoft.com/en-us/azure/devops/repos/tfvc/what-is-tfvc?view=azure-devops)
  • AccuRev (https://en.wikipedia.org/wiki/AccuRev_SCM)

Distributed Version Control Systems

See the live walkthrough.

• Conceptual Walkthrough - Follow this CS 262 deployment diagram example.
  1. Start with empty client GitHub repo (short for repository).
  2. “Copy” the repo down to a local machine to two local machines (A & B).
     • Distinguish different ways to “copy” the repo:
       • fork - new, independent application
       • clone - shared project - We’ll use clone.
     • Distinguish between Git and GitHub.
       • Git is the DVCS software.
       • GitHub is a cloud-based service that hosts Git repos.
     • Either way, the full copy is what makes it distributed.
       • Everyone has a full copy of the repo, including the full revision history.
       • Cf. maintaining/sharing a single, cloud-based copy shared by all.
  3. Client A add some code, commit it locally, and push the changes to GitHub.
     • New/modified files start as un-staged. add stages them for commit.
     • Distinguish different ways to “save” changes:
       • commit - save changes locally
       • push - save changes to GitHub
     • Note that not all files are staged/committed/pushed.
       • Some files are ignored (e.g., compiled code, temporary files, etc.).
  4. Client B add/commit changes locally but must pull changes before pushing, perhaps merge them with local changes.

References

• DVCS materials on Wednesday: https://github.com/kvlinden-courses/cs262/blob/master/materials/03management/class.md
• Git/GitHub deployment diagram: https://cs.calvin.edu/courses/cs/262/kvlinden/03management/images/dvcs.png
• Git commit/push/out (on quiz): https://i.ebayimg.com/images/g/GQQAAOSwYsBbxwgu/s-l640.jpg