Re: [GNUnet-developers] Decentralized database

[Christian Grothoff]

On 01/13/2018 09:00 PM, Amirouche Boubekki wrote:
> On 2018-01-13 20:23, Amirouche Boubekki wrote:
>>
>> Anyway, I figured that I can use the high level fs service to implement
>> a prototype. Basically, vertex will be files and links will be keywords.
>> That way, a new vertex can link to an existing vertex without having
>> to update somehow the old vertex.
>>
> 
> Here is a demo of it would work:
> https://asciinema.org/a/aFqvf54ABpb2DllB0jFhPbvKh
> 
> Simply said:
> 
> $ gnunet-publish vertex1
> $ gnunet-pbulish vertex2 -k $(sha256sum vertex1 | cut -d ' ' -f 1)
> $ gnunet-pbulish vertex3 -k $(sha256sum vertex1 | cut -d ' ' -f 1)
> $ gnunet-pbulish vertex4 -k $(sha256sum vertex1 | cut -d ' ' -f 1)
> 
> Then to know what links to vertex1:
> 
> $ gnunet-search $(sha256sum vertex1 | cut -d ' ' -f 1)
> 
> WDYT, it's viable somehow?

Well, the original idea was a bit different:

$ gnunet-publish -t v1 -P pseudonym -N v2  vertex1
$ gnunet-publish -t v2 -P pseudonym -N v3  vertex2
$ gnunet-publish -t v3 -P pseudonym -N v4  vertex3
$ gnunet-publish -t v4 -P pseudonym -N v5  vertex4

Basically, if you search for "v1" of "pseudonym", you find "vertex1"
with the promise that a future version will be published as "v2".
gnunet-search then *automatically* adds 'v2' to the search list, finds
"vertex2" (and then adds "v3" to the keyword list, etc.).

Updates must all be signed by the same "pseudonym". "v1" is just
arbitrary text (as long as it is unique), here chosen to illustrate the
point.

signature.asc
Description: OpenPGP digital signature