Yeah, we did that too: there was a disk file per message that described the mime parts, including mime type, and that information was available through the message store's file system interface. There was enough there to reconstruct the original message. This was a working system; we worked out the various issues.
- Re: [9fans] rfork(RFPROC) and ffork() erik quanstrom
- Re: [9fans] rfork(RFPROC) and ffork() geoff
- Re: [9fans] rfork(RFPROC) and ffork() Skip Tavakkolian
- Re: [9fans] rfork(RFPROC) and ffork() Lyndon Nerenberg
- Re: [9fans] rfork(RFPROC) and ffork() Dave Eckhardt
- Re: [9fans] rfork(RFPROC) and ffork() Russ Cox