-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Je Merkredo 18 Aprilo 2001 14:36, vi skribis:
> > 1) They are fairly static
>
> No, they are not at all static.
Perhaps we're talking about two different catalog mechanism. According to
the information on the web page (about how to make keys available), catalogs
are voluntarily published by the data holders, and are somehow propegated
through the system. My understanding is that each server has a catalog,
which can be fetched by request; for this to work, servers have to treat
catalogs as different from other data, so that catalogs aren't propegated
between servers -- if this happened, catalogs would become static. Since it
is a Bad Idea to start adding exceptions to a system (look at Java's
non-Object native types, the only things keeping Java from being a pure OO
platform), I assumed that this wouldn't be the case. However, your claims
that the catalogs are not static leads me to believe that you have indeed
started adding exceptions to the system.
> > 2) They are abusable
>
> All publically writable spaces are abusable. Therefore all unmoderated
> indexing and searching systems are abusable. Catalogs don't seem to be
> more abusable than other systems.
Again, my complaint only holds if catalogs aren't treated as key lists which
are unique to a specific server and non-propegated.
> > 3) The infrastructure to make catalogs work is much more complex than a
> > simple dynamic, distributed key-matching system, and provides very few
>
> It's less complex. And it's pretty much done.
Microsoft Windows is "pretty much done", too. Does that make it good or
efficient?
Humbug. You've added an entire mechanism which duplicates much of the work
of file sharing but with special behavior, for a single, specific case: the
transfer of catalogs, without propegation.
The non-propegation characteristic of catalogs reinforces my second primary
complaint about catalogs: the server contains an easily decoded list of names
and descriptions of the data stored on the server. This is information which
can be used in the prosecution of the server. Storing the keys as hashed
words is a one-way mechanism for matching keywords, which can not be used in
the prosecution of the server. Therefore, it is more secure.
The non-propegation characteristic of catalogs adds another two complaints:
Frist, this system will either increase network traffic, or have stasis.
Either each search by the user will cause a cascade of catalog requests, or
the catalogs will only be updated periodically, thereby rendering the
requestor's cached catalog static and out of date. You can alleviate this
problem by having a cache timestamp mechanism, but you're starting to add
complexity to the system that is required ONLY because you've chosen to use
catalogs. Second, this system is wasteful of resources. The client will
have some arbitrarily large cache of catalogs on their machine, however
temporarily, and I don't anticipate that the size of the catalogs will be
trivial if we assume even moderate popularity of Freenet.
=== SER Deutsch|Esperanto|Francaise|Linux|Java|Aikido|Dirigibles|GPG
=== http://www.germane-software.com/~ser jabber.com:ser ICQ:83578737
"One World, one Web, one Program" - Microsoft promotional ad
"Ein Volk, ein Reich, ein Fuhrer" - Adolf Hitler
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.2 (GNU/Linux)
Comment: For info see http://www.gnupg.org
iD8DBQE63xe/P0KxygnleI8RAjBPAKCtT3lztzNgLbbfg8axvOZBauMdAQCgk1pI
5g78Iw6nrHnkiFflGEKP+/U=
=SAGr
-----END PGP SIGNATURE-----
_______________________________________________
Devl mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/devl
