The idea is that your harddisk would be interpreted like a huge
TeXmacs document, or at least all TeXmacs files on your harddisk.
The TeXmacs file system would provide an interface to this "document"
for elementary operations, such as modifications in the text,
like replacing a subtree of the document by another one.
Of course, "the document" would still correspond to several physical
files on your harddisk, but the "TeXmacs file system" may organize
them in a way which is quite different from what is usually done.
In particular, it would maintain a large index for efficient search
operations. Possibly, it would also do some automatic version control.
My preferred solution would be server-based. In other words,
TeXmacs would constantly talk to some TeXmacs server which
would be in charge of maintaining your TeXmacs documents on
the hard disk. A server-based solution would have the additional
advantages that it can be developed independently from TeXmacs
itself and that it allows you to work in a remote fashion,
via the web.
I also recommend the use of Scheme, so that routines for
the manipulation of TeXmacs trees may be shared with those
in TeXmacs itself (at least in the future).
So: if you or anyone else is interested in creating such
a TeXmacs server, then that would be great. I don't expect
the task to be extremely difficult and new features could
be added incrementally. If you have some proposals or
further questions (like fixing precise protocols),
don't hesitate to post them to the list.