gnunet-developers
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [GNUnet-developers] End-user wishlist


From: Christian Grothoff
Subject: Re: [GNUnet-developers] End-user wishlist
Date: Tue, 24 Jun 2003 15:39:43 -0500
User-agent: KMail/1.4.3

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Tuesday 24 June 2003 01:46 pm, address@hidden wrote:
> This has been a quiet mailing list recently... maybe that has something to
> do with the fact that my GNUnet node is up and running stably for several
> days, doing what it's supposed to do, so there's nothing to complain about?
> Kudos to the bug-fixers. GNUnet's shaping up nicely.

Well, thank you :-). Another reason for it being more quiet here is that I've 
been distracted by other work for a while. Although I suspect that we'll have 
another minor bugfix release (0.5.4a) out rather soon. I'm currently working 
on implementing directories, so coding is still progressing.

> It must be a sign of legitimacy (for a very one-sided definition of
> legitimacy) that the new Harry Potter book was available on GNUnet less
> than 24 hours after its release. Now I'm dreaming about cool new features
> that I'd like to see built into and atop GNUnet, now that it's running
> stably on my machine:
>
> * A URL scheme for shared files, so they can be linked to from within
> hypertext documents

How about "gnunet://HASH1INHEX/HASH2INHEX/CRCINHEX/SIZE"? If someone writes a 
little wrapper for parsing/generating this type of URL, I'll put it in 
util/hashing.c to make it "official". Or are there any objections/problems 
with the above format?

> * A hypertext browser that makes use of GNUnet for transport and the above
> URL scheme (Please, no HTML though. This project is all about reinvention
> for robustness and learning from mistakes of the past--there's no good
> reason why one shouldn't abandon the cruft of the browser wars and embrace
> XML/XLink/XForms -> XSLT -> display.)

I think what you really want is a kio-plugin into konqueror or something like 
that. The plugin should parse GNUnet-directories (once implemented) and spit 
out whatever the browser can then understand (file-entries or HTML). But I'm 
not a KDE hacker, so it would probably be best if someone who is takes this 
on.

> * How's the namespace proposal coming? Has it reached concensus among
> developers yet? This one's still pretty "blue-sky" from an end-user
> perspective, but once GNUnet grows a bit, it will be critical for weeding
> out quality content from the spam and floody subversion attempts.

I think we have a concensus except for the hashCash/content rating via-voting 
amendment. The discussion died down a couple of weeks ago and I figured that 
Igor had some very good points at the end, which so far made me re-consider
this (for me, it basically boils down to that the assumption that the attacker 
has less CPU power than the honest participants is probably wrong, since the 
attacker might really just dedicate a machine to computing hash cash). So 
unless someone else comes up with some good arguments here, I guess we'll not 
have the hashCash content rating part (which leaves the system pretty much as 
proposed on the WWW).

> * Working chat facility! I've tried to get some folks to help me test it,
> but I understand the current chat tool is proof-of-concept only (and poor
> proof at that, since I've never seen it work). Anonymous chat would be
> perfect for discussing content found on GNUnet, sharing tastes, etc.

Oh, it works. Just start a gnunetd (make sure you load the chat module), open 
2 terminals, start gnunet-chat in each of them and type something. Now, you 
won't typically see messages from other users appear since hardly anyone is 
using it, but it's basic functionality is there. The major problem with the 
code is that the provisions against routing-loops are insufficient (if you 
have more than 20 or so chat messages in the network at a given time, 
messages may loop); that there are no chat-rooms; that all messages are 
broadcast in plaintext; that there is no protection for nick-names; that 
there is no 'sign on' or 'leave' and so on. I'm currently not too interested 
in improving this, but if you are, send patches :-).

> * Native front-end GUI tools for KDE, curses, other platforms, and a bit of
> design & improvement on the GTK+ frontend... for growing the user base, to
> include those who are scared of learning new interfaces.

There is gnunet-curses, http://www.ovmj.org/GNUnetCurses/, but development 
seems to have stopped for a while now -- and there's no link to the code or a 
contact address. Too bad. 

> Please don't take this list as a list of demands. I like GNUnet and
> evangelize it happily; I'm just voicing my dreams in hopes that some bored
> programmer will find them inspiring.

Sure, as Krista said, we're open for suggestions :-)

Christian
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQE++LcQ9tNtMeXQLkIRArT1AKCRbsPfhYJ+ou4T7UbTZwqMX3S9SwCfVJ66
+JOVxR5LNikNRA98mxRMTFY=
=GhpV
-----END PGP SIGNATURE-----





reply via email to

[Prev in Thread] Current Thread [Next in Thread]