[no_more_death]

1. Browser computes hash for all JS files loaded, and tallies number of uses.

2. JS files that get a certain number of uses get marked for compilation.

3. On browser startup, the marked files are recompiled into the binary.

4. The modules are pulled in when the browser detects an HTTP requested file matches an existing file.

MD5 hashing couldn't be used, due to collision vulnerabilities!

I guess the question is whether the browser's JITter is already as fast as compiling javascript into the binary.