[Nix-dev] [***SPAM***] Re: Google Summer of Code 2015

[Michael Raskin]

>> This is called "breaking many use cases for email". This means I cannot
>> write email on a blackboard in the beginning of a talk.
>
>Not so, I issue an Interest packet which floods into the network and
>finds you. The Interest packet says Michael, Stewart has an email for
>you, please pull it.
>You issue an Interest and which floods the network to me and I respond.

And how deciding to pull or not to pull this email is different from
deciding to read or not read a email now?

>> For name resolution you either need a lot of address conversion
>> infrastructure or to flood everyone at some points.
>
>There is no name resolution, The Interest packet is flooded into the
>network, each node checks to see if they have data which matches the
>Interest packet.
>Names and Locations are divorced.

I am not sure that this flood of named requests is better than
suboptimal delivery of popular content.

>> It is a big deal once your content request becomes LRU-forgotten before
>> the reply arrives.
>
>If there is no pending Interests then the data is dropped immediately,
>fear not, the data will pulled another way.

I need to have large timeout on all routers along at least one path.
Percolation theory says it is not a trivial question…

>> Routers are likely to have their own certified list and only allow you
>> to see things they have in their trust system.
>Could be, any different from a database?

Yes, because again I need an unbroken chain.

>> None of NDN benefits happen unless you actually route something via
>> caching routers.
>Yes buffers have been used as part of networking for decades now. It's
>not new just the buffers can be a database,
>a filesystem, a memory stick whatever.

The size of the buffers and characteristic times are different. The 
degradation also happens differently.

>> When you use /nix/store/ as cache, it is cheaper to do properly
>> checksummed P2P over TCP/IP and not help a technology likely to make
>> Internet filtering cheaper.
>
>What is cheaper? Secondly how do you cheaply incorporate a temporal
>binding? Not going to happen.

When you have no in-transfer caching, what is the benefit of NDN?

Freenet does caching of popular content, too. Bittorrent also 
effectively does.

NDN makes filtering cheaper, because the request for specific content
is easier to distinguish. Also it makes it creates incentives for 
routers to ignore non-popular content.

>> Now, collecting unreferenced old paths may be a good idea. Removing
>> rarely-used still-referenced-from-profile paths is malicious.
>>
>> Some links are slow; some rarely used things are important when they
>> are used.
>
>Yes you are correct it is malicious if you have no network connection,
>then you could argue being disconnected from the Internet is
>malicious.

Not having a reasonable connection to the Internet some of the time is
a norm in most places, US included.

>Which I agree with!

NDN helps effectively disconnecting people from Internet, in my opinion.