106 lines
5.4 KiB
Plaintext
106 lines
5.4 KiB
Plaintext
An Alpha release of version 5.2 of the Prospero file system is now
|
|
available by anonymous FTP from PROSPERO.ISI.EDU in the file
|
|
pub/prospero/prospero-alpha.5.2.tar.Z. It can also be obtained
|
|
through Prospero itself from /releases/prospero/prospero-alpha.5.2.tar.Z.
|
|
The name prospero.tar.Z in those directories still points to the stable
|
|
Beta 5.1 release; this has been done for the sake of the archie server
|
|
maintainers and other information providers who depend upon Prospero.
|
|
|
|
Some important changes since the last Prospero release should be of
|
|
interest to the readers of comp.infosystems.gopher:
|
|
|
|
(a) A menu-based browser client, similar to the Gopher client
|
|
(b) A gateway server which makes all of Gopherspace available through
|
|
Prospero queries.
|
|
(c) A menu-based browser API interface library to Prospero, which allows
|
|
one to easily write new browsers that use Prospero to make their underlying
|
|
calls.
|
|
|
|
The menu-based browser supports all of the basic Gopher types:
|
|
searches, text files, data files, and portals (what Gopher calls
|
|
telnet sessions). It does not yet support sophisticated display of
|
|
binary image files, reading Postscript or MIME documents, playing
|
|
sounds, or automatic decoding of uuencoded, compressed, or tarred
|
|
files. These enhancements will be added shortly; the protocol
|
|
definitions for them have already been made.
|
|
|
|
We support authenticated remote modifications to menus, though not
|
|
presently from within the browser itself. For now you will have to
|
|
learn how to use a few Prospero commands to add new menus or items and
|
|
set descriptions. We plan to eventually support remote update within
|
|
the browser itself.
|
|
|
|
Prospero, and thus menus available to the menu browser (as a Prospero
|
|
client), supports fine grained access control based on client
|
|
identity. Authentication is provided using 4 alternative
|
|
authentication mechanisms, depending on your specific needs: (a)
|
|
Kerberos (version 5); (b) per user passwords; (c) trusted port and
|
|
trusted host assertion (like the Berkeley R commands); and (d) a check
|
|
of the host from which a request originated and the asserted user
|
|
identity of the client.
|
|
|
|
Prospero provides full access control list mechanisms for those who
|
|
wish to put fine-grained permissions on particular objects or
|
|
directories. For instance, one can easily set some directories so
|
|
that anybody at your company or university can read data in them, only
|
|
a particular group of people can insert data into them, and only the
|
|
system administrator or the author of a particular piece of
|
|
information is allowed to modify and delete it. Of course, anyone who
|
|
does not need these security mechanisms can export data using the
|
|
default permissions defined on a per server basis.
|
|
|
|
An elegant new Prospero feature we're rather proud of is the search
|
|
interface. We support a variety of search methods, including direct
|
|
queries to Prospero-speaking database servers, such as Archie servers
|
|
and all Gopher servers via a Prospero-Gopher gateway. The search
|
|
servers are self defining, exporting attributes that tell the client
|
|
what to prompt for, help messages to display if requested by the user,
|
|
and how to package the search when it is sent to the server.
|
|
|
|
Client implementors need not worry about the generality of this
|
|
mechanism since they can make perfectly functional clients that use
|
|
only a small subset of the available data. The menu browser has a
|
|
subset of this search interface built into it.
|
|
|
|
The philosophy behind this work, or why we are doing this:
|
|
|
|
One of Gopher's greatest strengths is its clear, straightforward user
|
|
interface. However, the gopher protocol itself has a number of
|
|
limitations which keep it from being suitable as a universal Internet
|
|
information infrastructure, such as the fact that all updates to data
|
|
must be made through means outside the protocol (i.e., editing files
|
|
on the server) and its lack of security mechanism.
|
|
|
|
Most current Internet information services, such as Gopher, World Wide
|
|
Web, and WAIS, have their own specialized clients, servers, and
|
|
network protocols. This makes it difficult for these services to
|
|
interoperate smoothly. People attempt to address these issues with
|
|
inefficient ad-hoc gateways, such as the Gopher servers that make
|
|
Archie queries on behalf of Gopher clients, or even our own
|
|
Prospero-Gopher gateway. These are not very good solutions.
|
|
|
|
Prospero provides a framework for integrating information from diverse
|
|
sources. Client writers can work on interesting clients, and service
|
|
providers can publish their data, without needing to work on issues of
|
|
little interest to them. We explore these ideas in greater depth in
|
|
the paper 'Prospero: A Base for Building Information Infrastructure',
|
|
to be presented at the INET '93 conference in August, and available
|
|
via anonymous FTP from PROSPERO.ISI.EDU as
|
|
/pub/papers/prospero/prospero-bii.ps.Z.
|
|
|
|
We encourage others to build clients that make their queries through
|
|
Prospero and to publish their data through Prospero. We provide an
|
|
application library (much better documented in this release than in
|
|
previous ones) and a simple application interface designed
|
|
specifically for the authors of read-only menu browsers. It is also
|
|
not difficult to publish your data through Prospero, and we are eager
|
|
to work with those who wish to do so.
|
|
|
|
-- Steven Seger Augart
|
|
Research Staff, USC Information Sciences Institute
|
|
|
|
-- B. Clifford Neuman
|
|
Scientist, USC Information Sciences Institute
|
|
|
|
email: info-prospero@isi.edu
|