Re: [GNUnet-developers] packaging GNUnet for OpenWrt

[Christian Grothoff]

I'm putting this back on the list, as we're no longer really discussing
subtle clarifications...

On 06/05/2015 03:05 PM, Daniel Golle wrote:
> Hi Christian,
> 
>> So you could go for datacache=heap, namestore=flat file (to be
>> implemented) and peerstore = stub/heap --- and disable datastore.
> 
> Yes, this sounds like what could a sane minimal supported setup on
> small devices.
> How far is GNUnet from providing the stub/heap/static file-based
> implementations for namestore and peerstore?
> How much effort do you estimate it would take to implement the missing
> parts?

Look at
https://gnunet.org/svn/gnunet/src/namestore/plugin_namestore_sqlite.c
and

https://gnunet.org/svn/gnunet/src/peerstore/plugin_peerstore_sqlite.c

About 500 lines of actual C code each, basically insert, retrieve,
delete functions.  I don't expect a heap/file-backed version to be
significantly harder to write -- we have a hashmap implementation in
util/, so it's really just keeping it all on the heap and serializing
to/from disk.  Skilled dev can probably do either in a day or two.

>>> Another question: There are still many systems without a
>>> name-service-switch and thus no way to configure /etc/nsswitch.conf to
>>> try resolution via GNS. I guess I can still use dns2gns and have
>>> dnsmasq delegate some or all TLDs to the local dns2gns instance, right?
>>>
>>
>> Yes, you can use dns2gns, or the gnunet-gns-proxy (socks proxy for
>> HTTP), or gnunet-service-dns (which can use iptables to redirect DNS
>> traffic to a virtual interface), or of course some apps might support
>> the GNS protocol "natively".  Many ways of doing it, NSS is just one.
> 
> Cool, so lack of NSS isn't a show-stopper :)

Yes, absolutely not.  NSS is good if you want personalized resolution
based on UID, but I don't see that as useful for OpenWRT anyway.

Happy hacking!

Christian

0xE29FC3CC.asc
Description: application/pgp-keys

signature.asc
Description: OpenPGP digital signature