This is a bare-bones implementation of what's described in MIT AI Memo 1564. There's nothing of substance here; all the heavy lifting is done by scheme 48. So if you want to know what W7 is, the answer is that it's an "extended subset" of Scheme 48: Scheme 48 sanitized for capability security, with a bit of miscellaneous shim added.
This code is mostly untested and probably both incomplete and wrong.
There are a number of embarrassments, requiring a number of apologies.
- No secure interface is provided: not via TCP/IP, command line, or any other mode. From the Scheme 48 command line, you still have access to the configuration environment via comma-commands, from which you can do anything. This could be repaired by creating a restricted configuration environment - Scheme 48 is certainly capable of doing that.
- Reification uses Scheme 48 configuration environment reification (config-package), which seems an awful kludge. There should be a more direct way to do it that doesn't haul in so much of Scheme 48's programming environment infrastructure.
- There is an HTTP server module, but it is disabled because it depends on some HTTP client and server code of mine that is really substandard. And as it says in the report there are no distributed capabilities. (See the Psyche repository if you must.)
- 'sharp-sharp' is a global resource that is not properly managed.
To play around with it:
- Start up a Scheme 48 command processor
,config ,load w7-config.scm,config (define-structure scratch (export) (open usual-w7-features)),in scratch
Welcome to jail!