Re: [Help-gnunet] Information request in preparation for a talk

[Christian Grothoff]

On 04/24/2015 08:05 AM, Antoine Guellier wrote:
> Dear GNUnet contributors/developers,
> 
> I am a PhD student at Inria Rennes (France), and I come here to request
> more information on GNUnet. Indeed, I am to make a small talk in a week
> or so on the subject of my choosing at Deakin Uni (Australia) and I
> would like to check my understanding of the GNUnet project as a whole.
> I've been through all the doc I could find without finding definite
> answers to some of my questions:
> 
> - I tried to draw an ISO-like stack of the networking components in
> GNUnet (see attachments). Could you (in)validate my understanding? Feel
> free to modify the LaTeX source.

You might want to look at svn/gnunet/doc/structure.dot (I generated the
attached PS from that using "dot" from Graphviz).   It's not 100%
complete, but probably close to what you are looking for.  And we don't
quite follow "ISO" ;-).

> - Do the modules in the src/ folder correspond to each block/layer in
> the diagram? e.g. src/core for CORE, src/cadet for CADET

Yes.

> - Where are the anonymity and privacy ensured? Apparently, only the
> file-sharing application can do anonymity. But new  components, such as
> an anonymous version of R5N, could be developed, couldn't they?

Yes, and we've started with that in src/rps/, but it's not
finished/working yet.

> - What DHT/routing is actually in the src/dht module? R5N, GAP, X-Vine?

R5N and X-vine, there are the sources with "x" for X-vine, and those
without for R5N.  GAP is well-hidden inside of src/fs/.

> - What is the src/mesh module? Another name for CADET, i.e. a way to
> indirectly reach arbitrary peers in the network via a multi-hop path?

MESH is the old name for CADET. src/mesh/ should be dead/gone for a
while now, but outdated docs may still reference it. And yes, you're
right about the purpose.

> Finally, I want to express my support to the GNUnet contributors. I
> believe it is a very interesting (even necessary) project, and its goals
> are laudable! It includes many good ideas (the public key oriented
> routing, work  on anything from Ethernet to HTTP, and automatic
> transport selection). In my thesis, I work on privacy in networking
> using homomorphic cryptography, and I find your approach of redefining
> the whole protocol stack "courageous" and very welcome.
> However, although the large documentation on the GNUnet project is
> welcome, navigating between the various sources (doxygen, the FAQ, the
> developper handbook, the papers, ...) feels like a maze, especially
> because some documentation is outdated. But I understand that
> maintaining doc is laborious!

True, if you find any documentation that is outdated, I appreciate
pointers -- via e-mail or by filing bugs on Mantis. Because even just
finding the outdated parts is laborious ;-).

-Christian

structure.ps
Description: PostScript document

0xE29FC3CC.asc
Description: application/pgp-keys