This directory includes the preceding backend evaluation and version 1.4 release checks.
The original benchmark and browser JSON records are retained as measured.
Version 1.4 changes only product labels in the player HTML, not its JavaScript/CSS.
See v1_4_release.json and ../VALIDATION_1_4.txt for current kit checks.

Validation evidence for the PDFium prototype

Run Python regressions from the kit root:
  python -m unittest discover -s tests -v
Only requirements.txt is needed for these tests.

compare_decks.py compares old/new deck structure and selected vector handout pages.
benchmark_convert.py runs one conversion using the chosen source kit and Python.
The original source archives are available at the URLs in REFERENCE_SOURCES.json.
They are not duplicated in this package. Use separate environments for old/new PDF backends.

browser_validate.cjs is optional developer tooling requiring Node.js and Playwright.
It is NOT a runtime dependency of SCVeamer. Set SCVEAMER_BROWSER_EXECUTABLE when
using an existing compatible Chromium executable, or use Playwright's installed one.
The script opens local HTML with an offline browser context. It records process
PSS/RSS on Linux and keeps JS heap measurements separate.

JSON records preserve measured results; environment-local paths are normalized.
The image-cache experiment was not adopted. It is evidence, not active player code.
The earlier rejected pypdf full-text strategy is not included as a successful result.
Temporary browser screenshots are not included.

The two comparison PNGs are extracted/rendered source-page details, not browser
screenshots. SVG was rendered using librsvg; PNG resampling used Lanczos.
