Re: A broken brand?
Sandro Magi <[email protected]>
| Newsgroups | gmane.comp.lang.e.general |
|---|---|
| Message-ID | <[email protected]> |
David Wagner wrote: > Sandro writes: > >> MarkM clarified in a follow-up: types wouldn't have helped if you want >> to retain transparent intermediation. >> > > Ahh, my apologies for missing that. Can you explain why we'd want > transparent intermediation? Are we talking about the membrane pattern, > or something like that? Unless I'm mistaken, this is a problem for sealers, which have many uses that cross machine boundaries. I should amend my original statement to: types wouldn't help if you want to retain intermediation given currently popular statically typed languages. There are prototypes for statically typed distributed languages, like HashCaml [1], and Alice ML [2] which maintain type safety across machine boundaries. Types don't mean much in a distributed application where the runtimes can't trust each other, not if the runtime hasn't been built with this use in mind. Distributed typing based on secrets can probably enforce the necessary invariants. Sandro [1] http://www.cl.cam.ac.uk/~pes20/hashcaml/ [2] http://www.ps.uni-sb.de/alice/