Evaluate upstream packet-table sizing, Full Names and path popup improvements (2090) #217

Deferred evaluation

Assess the useful parts of upstream PR 2090, merged on 2026-10-03. This is a backlog assessment, not approval to implement or merge a port.

Comparison snapshot: our master 8bafcdf2a6bb0014127bce4822c8ad0399bd2f12, upstream master 9e49db9725d6346959c41ec23c66d96b7e43cc4c, checked 2026-10-04.

What it does / potential benefit

• Fits short packet columns to their content, leaving the remaining space for Path and Details; avoids treating a virtual-scroll spacer's colspan as a real column width.
• Adds a Full Names toggle, a saved browser preference and a URL override.
• Improves the full-path overflow popup, including hop counts, hover/focus, keyboard activation and dismissal.
• Uses one correctly oriented expand arrow rather than overlapping indicators.

Potential value: more readable packet rows and long node names, with fewer manual width adjustments. It does not improve packet decoding or make hop identities more certain.

What we already have

Our packets.js already has resizable columns, path truncation, a +N pill and a click-open full-path popup. The new feature would be an enhancement, not a first implementation.

The Full Names toggle and upstream's new sizing helpers are absent from the checked master. The collapsed up-caret is already tracked in #189; do not create a second independent fix for it.

Open #185 changes hop wrappers, warning summaries and popup counting. #211 and #212 also touch packet UI code. Recheck their state and integrate against the current master before selecting any upstream hunks.

Risks / decisions for later

• Preserve grouped/flat rows, observation selection, saved widths, URL parameters and narrow-screen behavior.
• Count actual hops, not arrows, wrappers or ambiguity warnings.
• Upstream performance results are not measurements of our fork. Width measurement must stay bounded to rendered rows and must not add expensive work on every WebSocket event.
• Decide whether to adopt the full interaction bundle or only sizing and Full Names. Avoid silently discarding saved user widths without a deliberate compatibility decision.

Evaluation checklist

• Compare with feat(packets): show hop ambiguity, resolved per observer (port of upstream 2099) #185 and test: 13 red orphan unit tests + 4 orphan E2E files left after #187; collapsed packet groups show an up-caret #189; list only the remaining gaps.
• Test virtual-scroll spacers, resize, column hiding, delayed observer names and persisted preferences.
• Validate long paths, keyboard/focus behavior and popup cleanup in desktop/mobile browsers.
• Measure initial rendering and scrolling on a realistic large packet set.
• Record a decision: adapt selected parts, adopt all, or retain our implementation. No implementation is required merely to close the assessment.