There are many places in the codebase that call our implementation of https://tc39.es/ecma262/#sec-parse-script

• FuzzilJs.cpp
• FuzzJs.cpp
• js_repl.cpp (WASM runtime)
• test-bytecode-js.cpp
• Spreadsheet.cpp
• $262Object.cpp
• JavaScriptTestRunner.h
• DOM/Document.cpp
• HTML/Scripting/ClassicScript.cpp
• js.cpp

Approximately zero of these places validate that the input is valid UTF-8.

However, the ECMA262 spec says that the "source text" of ParseScript ( sourceText, realm, hostDefined ) must be ECMAScript Source Text (https://tc39.es/ecma262/#sec-source-text), which is mandated to be valid UTF-8.

So onto the question: Whose job is it to mandate that the script you're passing to ParseScript be UTF-8? The caller? Or can the algorithm look at the input and say "ew, that's not UTF-8, here's a parse error".

Certanitly we shouldn't VERIFY() and crash in a String::from_deprecated_string(source_text).release_value_but_fixme_should_propagate_errors in ParseText(sourceText, Script), like we currently do...