"signed scripts are a total failure"

What does signing a script actually do for the end user?

anyone can sign a script, it just takes lots of effort. Once a script is signed, it still has to follow rules setup by the user/user-agent. It can either be allowed to do dangerous/annoying things or not allowed to do them. All signing scripts does is require extra effort both on the part of the content developer and the user.

I don't remember what the current permissions dialogs look like, but i can say a few things about them.

  1. They're confusing.
  2. They don't say much.
  3. They aren't memorable.
  4. The options they provided weren't useful.

"a useful dialog"

what we should have is the n4 style java permissions dialog

any script can try to use a privilege or ask for a privilege the more polite it is the friendlier the dialog will appear.

so if it's a signed script and it asked for permission to do something, the dialog will say so

if it's a signed script that tried to do something without asking, the dialog would say so

if it's an unsigned script that asked, it'd say so

..

the dialog would explain what the permission could be used to do and would allow remember for .. as well as accept/reject/ignore (ignore=throw no errors but don't actually do whatever thing the script wanted to do)

"a running script"

Any time the current page has a script running, there would be an icon in the status bar the icon would give some indication about it being busy, having privileges, .. and clicking it would enable you to change the settings for the script, including stopping it.

A site which had a <script> tag but where you configured scripts not to run would have a script icon w/ a universal no icon (probably a slash). This would enable you to recognize that a script was available and that it wasn't running. You could then click the icon, change the settings and let the scripts run for the session.

For the most part the settings you could twiddle in the script dialog would be the universal ones (can run at all, can twiddle the window object, can twiddle the navigator object), plus whatever things the script has asked about or tried to use, or has been previously configured not to do there /might/ be an advanced way to add a rule for something which wasn't currently in those lists, but there shouldn't be a need w.r.t. the app actually tossing up a dialog about permissions, it might not actually throw one up, but might instead flash the script icon w/ a dialog icon ('i' thought bubble) that way it's the user's choice to be annoyed by a permissions dialog, instead of the script's or the browser's.