# IFVG Desk 1.2.0 — website delivery verification

This release adds a bounded setup journal and diagnostic export. It does not add a new entry model. The journal saves reasons on confirmed bars; paper acceptance is not a broker fill.

## Files and installation

- [Current two-script bundle](ifvg-desk-v1.2.0-web.1.zip), [checksum](ifvg-desk-v1.2.0-web.1.zip.sha256), and [display patch notes](v1.2.0-web.1/PATCH-NOTES.md).
- [Complete original release ZIP](ifvg-desk-v1.2.0.zip) and [original SHA-256 checksum](ifvg-desk-v1.2.0.zip.sha256).
- [Indicator](ifvg-desk.pine), [Strategy Lab](ifvg-desk-strategy.pine), and [MIT license](LICENSE).
- [Quick start](v1.2.0/docs/quick-start.md), [setup history and export guide](v1.2.0/docs/audit-trail-v1.2.md), and [bug report template](v1.2.0/docs/bug-report-template.md).
- [Original manifest](v1.2.0/MANIFEST.json) and [website release metadata](release.json).

The original ZIP contains 98 manifest-listed payloads plus the manifest. Every payload and the complete website mirror match the original bytes. Rebuilding the original scripts, audit probe and product-only package independently reproduced all 98 payloads and the manifest byte for byte.

The current website scripts use **web revision display-fix.1**: three corrected display lines plus one identifying header comment. They clarify skipped inversion entries without presenting current checks as historical reasons, and label the paper count “closed positive.” Trading rules and the order adapter are unchanged. The copyable TypeScript constants exactly match these corrected downloads. The current small bundle has its own manifest; it must not be confused with the original development ZIP.

## Checks and their scope

The website import independently reran all **88 local checks** from the immutable package in an isolated copy: [outcome and source hashes](v1.2.0-local-checks.json), [actual TAP output](v1.2.0-local-checks.tap). These are JavaScript specification fixtures, Pine source guards and delivery-tool tests, not Pine execution.

The producer supplied native evidence for three successful compilations and nine passing audit-probe rows on a standard MNQ 5-minute chart: [compiler results](v1.2.0/docs/evidence/v1.2.0/compile.json), [probe rows](v1.2.0/docs/evidence/v1.2.0/runtime-probe.json), [source hashes](v1.2.0/docs/evidence/v1.2.0/source-hashes.json), and [probe capture](v1.2.0/docs/screenshots/v1.2.0/audit-probe.png). This evidence identifies the original package sources. Native runtime probes were not rerun for this website import; separate web-revision compilation is recorded below and in the release metadata.

Both corrected website scripts also passed a fresh TradingView native compilation with zero errors and zero warnings: [web-revision compiler results and exact source hashes](v1.2.0-web-compile.json). Compilation checks the two served sources; it does not repeat the nine native audit fixtures, chart replay or broker simulation.

The producer's [before/after comparison](v1.2.0/docs/evidence/v1.2.0/equivalence.json) uses the same 5,888 loaded historical MNQ 5-minute bars and feed boundaries for both entry modes. Its matching decision checksums, level/size totals and paper state are bounded regression evidence. This is not proof of every input combination, every market, live/reload equivalence or broker fills. The strategy broker backtest was not repeated for this diagnostic revision.

The [setup history capture](v1.2.0/docs/screenshots/v1.2.0/setup-history.png) and [saved panel rows](v1.2.0/docs/evidence/v1.2.0/history-panel.json) show an actual chart with the optional panel and Pine Logs export enabled. Those are the preview's only two default overrides; its reconstructed hash matches the producer record. The [report sample](v1.2.0/docs/evidence/v1.2.0/report-sample.json) contains a settings header and six representative event objects, not a full export of every retained record.

## Limits that matter

Recording defaults on. The history panel and Pine Logs export default off. Retention defaults to 300 events, with a maximum of 500 across all setups. Older events can be removed; a filtered setup can therefore have missing earlier events. Chart or input changes rebuild the journal from available history. This is not permanent storage or an exhaustive broker journal.

The two earlier website captures remain historical v1.1.1 images. Earlier measurements in the bundled documentation are historical research, not a new v1.2.0 broker test. No trading edge, future return or protected-indicator parity is established. Read the [producer verification report](v1.2.0/docs/verification-v1.2.md) for its complete scope.