This hierarchical filesystems sounds very good, files need not to
be stored under their real names or even on the same system, maybe they
even could on a IBM's File Storage Facility for which I haven't found any
use yet :-)  Files sent to special list e.g. REXXSRC would be forwarded
to the editor for inspection and put on the REXXSRC filelist or it's
sublist by the 'editor' of that program section on fileserver and
distributed to all peer REXXSRC lists as well. Voila, we have a
REAL distributed fileserver network to rival GRAND et al as a byproduct
of a 'little mailing list expander' ;-)
   Other feature I'd still like to have is automatical index of subjects or
first words of the letters text which I now do with my modifications
to BITNIC's old listserver for file archival. That listarc has about 40-50
lists but NO subscribers anymore and will disappear soon totally, I hope.
I'd like to store all letters as single files and be able to get them
by looking at the description field. Individual storage would also
help integration with netnews and reduce network load when retrieving archives.
Larger archives could be arranged using package files for every month,
Volume, you name it. I've received the manuals of the Princeton NetNews system
and it might be an alternative way to read lists than plain MAIL/MAILBOOK
besides being UseNet compatible news system.
 
Ps. Eric: Yesterday there was 46 reguests on the AMIGO proposal and they still
argued on QZ-COM how and who should check the final version :-)