perfgate-render

Human-readable output from performance comparisons.

Takes a CompareReceipt and turns it into something a human can read -- a Markdown table for a PR comment, GitHub Actions annotations for the checks tab, or a fully custom layout via Handlebars templates.

Output Formats

Function	Output	Where it shows up
`render_markdown`	Markdown table with verdicts	PR comments, CI summaries
`github_annotations`	`::error::` / `::warning::`	GitHub Actions checks tab
`render_markdown_template`	Custom Handlebars template	Anywhere you need a custom layout

Quick Start

use perfgate_render::{render_markdown, github_annotations, render_markdown_template};

// Markdown table for a PR comment
let md = render_markdown(&compare_receipt);

// GitHub Actions annotations (only emitted for warn/fail metrics)
let annotations = github_annotations(&compare_receipt);

// Custom template
let tmpl = "{{header}}\n{{#each rows}}- {{metric}}: {{delta_pct}}\n{{/each}}";
let custom = render_markdown_template(&compare_receipt, tmpl)?;

Helpers

Formatting functions used internally, also exported for custom renderers:

Function	Example output
`format_value(metric, v)`	`"42"`, `"1.234"`
`format_pct(pct)`	`"+10.00%"`, `"-3.50%"`
`direction_str(dir)`	`"lower"`, `"higher"`
`metric_status_icon(status)`	pass / warn / fail / skip
`metric_status_str(status)`	`"pass"`, `"warn"`
`parse_reason_token(token)`	`"wall_ms_warn"` -> `(WallMs, Warn)`

Template Context

render_markdown_template exposes a JSON context with header, bench, verdict, rows (array of per-metric objects with metric, delta_pct, status_icon, raw, etc.), reasons, and the full compare receipt.

License