Skip to content

feat: DAG view expandables implementation and improved task identification view with CTAs from DAG#180

Open
Vinyl-Davyl wants to merge 8 commits intoNetflix:masterfrom
Vinyl-Davyl:feat/improved-task-identification-ui-and-better-foreach-identification
Open

feat: DAG view expandables implementation and improved task identification view with CTAs from DAG#180
Vinyl-Davyl wants to merge 8 commits intoNetflix:masterfrom
Vinyl-Davyl:feat/improved-task-identification-ui-and-better-foreach-identification

Conversation

@Vinyl-Davyl
Copy link
Copy Markdown

Requirements for a pull request

  • Unit tests related to the change have been updated
  • Documentation related to the change has been updated

Description of the Change

Closes Issue #88 and Partially #89

This PR introduces high-density monitoring capabilities to the DAG visualization while strictly adhering to the project's design system.

  1. "Mission Control" Expanded DAG Mode:

Added a new "Expand DAG" toggle. Steps now expand to show a high-density grid of task iterations.
Each iteration card indicates status via color-coding and surfaces iteration values (from foreach-stack metadata) directly.
Full point-and-click navigation from any graph card to its respective task log.

  1. Clean Layout Logic:

Switched graph connections to a discrete VerticalConnector system. This architectural change ensures connection lines only exist between nodes, physically eliminating bleed-through artifacts into boxes or containers.

  1. Theme-First Styler:

Centralized all background and status colors into
theme.css
.
Removed all hardcoded hex values partaining to DAG (#f4faf4, #ffffff, etc.) in favor of standard CSS variables like var(--color-success-bg) for future Dark Mode Implementation prep.

  1. Optimized Timeline Labels:

Replaced technical internal IDs with user-friendly, inline value labels (TaskID · Value).
Automatically strips system prefixes for a cleaner, non-technical scanning experience.

Alternate Designs

Background Layering: Previously used a single background line, but it caused visual artifacts in complex foreach layers. The current connector system is graphically superior.

Possible Drawbacks

Wide foreaches will create deep nodes; this is managed by the graph's existing zoom-to-fit and scrolling mechanisms.

Verification Process

Run project.
Open the DAG tab and toggle "Expand DAG".
Verify that the task grid tiles have subtle, theme-consistent background colors.
Click on one of the expanded item to redirect to the task(s) open
Verify that the vertical lines are perfectly contained between nodes with zero bleed into the white or gray boxes.

Release Notes

Added an "Expand DAG" mode for granular task monitoring and update task identification flow from DAG, refactored the DAG component to use centralized theme variables for all status and structural colors.

Screenshot 2026-03-16 at 8 57 52 AM Screenshot 2026-03-16 at 8 58 29 AM Screenshot 2026-03-16 at 8 58 53 AM Screenshot 2026-03-16 at 8 59 21 AM

…etflix#88)

Show a small pill badge below task IDs in the timeline for tasks that
are part of a foreach loop. The badge displays the foreach variable
value (e.g. 'country:US'), making it trivially easy to identify which
task corresponds to which iteration without opening each one.

Changes:
- TaskListLabel.tsx: render ForeachLabel badge when task.foreach_label
  is set; update tooltip title to include the value
- en.ts: add 'task.foreach-label' translation key
- browser.js (mocks): add foreach_label to sample tasks for dev testing
@Vinyl-Davyl Vinyl-Davyl changed the title feat: DAG view expandables implementation and improved task identification view CTAs from DAG feat: DAG view expandables implementation and improved task identification view with CTAs from DAG Mar 16, 2026
@Vinyl-Davyl
Copy link
Copy Markdown
Author

cc: @saikonen @romain-intel

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant