Real spreadsheets and the triple machine

Spreadsheets make great interfaces to the machine, they agree on int vs text, the machine adds serialized data format. My approach is to use the Open Office spreadsheet javascript package and make an unpacker/packer for nested form. Hence, the select of dot and comma, they make the nested order look like a simple extension of normal csv format, just allowing variable length rows and nesting.

On layout, I just indent for each descent step, and it lays out naturally. Great for accounting, the ability to load a container of int that relates to some string key word. Or use nesting to compute sub totals and make then available for search.

Which brings up the machine and word processors. Do what you want with nested order and the text blob. Pick constant keywords, like PARAGAPH, CHAPTER and so on. Rig the local scripter to use TE grammar on output, and put the documents in triple store. The client can ope up a little TE window and use it to track the results: (CurrentDocument:
(Chapters.*.headings)) (track the ongoing list of chapters, collect everything o the chapters from the start to and including heading. The little search engine could keep a display of ongoing searches.

More importantly, when your word processor or spread sheet is talking to a remote machine, the thing still works, no browser, at least in version 3. In that version the URL attribute will be implemented, remote table references become transparent. In version 4, the industry will get it, and have a specific web method: triple://www.SomeTriplePlace.com