Investigate innerHTML Reproducibility

[deckar01]

I suspect that document.documentElement.innerHTML may not produce a deterministic value across browsers and versions for all HTML content.

Consider:

• Normalizing the HTML string
• Using the original HTTP response body before DOM processing

[deckar01]

HTML file → DOM → innerHTML

There are 2 operations that could differ across browsers. The complex parsing operation from an HTML file to DOM elements, and the serialization of innerHTML from DOM elements to a string.

Neither of these operations are guaranteed to behave the same across browsers and should not be relied on when generating the info hash. The info hash should describe the original HTML file contents without any transformation.

The client's scripting environment does not have direct access to this content as a string, but it can be fetched as an AJAX request. Assuming the HTML content is served with the proper cache headers, an AJAX request for the original HTML file should return the original file content as a string from the browser's cache without fetching it over HTTP again.

HTML file → Browser cache → AJAX request

[deckar01]

Fixed in 95ff4d9.