> Also note that, if I specify my goal as viewing a PDF document, I am explicitly telling the computer I that running a script is not something I want to be doi

> Also note that, if I specify my goal as viewing a PDF document, I am explicitly telling the computer I that running a script is not something I want to be doing. LOL, no, you don't. At least in general, for interacting with rich documents. Imagine for a moment disabling JavaScript in your browser and browsing HN. Or any other important website. Now go ahead and actually try that -- you'll find that you've vastly underestimated the impact on usability. That you've missed a lot of functionality that is seamless, unobtrusive -- and you only notice its importance once its gone. Thank you, I rest my case [0] [1]. > (...) you'd specify its permissions ahead of time, in the form of a natural-language phrase that is fuzzily-keyed in a HTM memory-base (...) It strikes me as something very open to attacking special cases. Like weak encryption -- works most of the time, but breakable by somebody with incentives. But that's just a hitch, no hard facts. The UX reeks of the UAC -- I can't see any bottom-line minding company repeating that mistake. ---- [0] yes, I know some people use the `noscript' plugin or similar. Still, the second basic functionality of such plugin is whitelisting websites for enabling JS on 'em. [1] In any case, a bug in interpreter of any no-scripted (`plain data') document can give way to memory corruption and code injection. Happened way too many times with browsers, movie players etc. Exploit hidden in PNG? Heck, why not?