Plugin interface about namecoin-legacy HOT 5 CLOSED

I think the best way to integrate with namecoind is to use the RPC interface. Do you have concrete examples of things which could be implemented via RPC only with a significant performance overhead (or other problems)? All of the things you mentioned above seem suited to base off the RPC interface. IMHO, the separation into multiple processes and communication over RPC seems like a good choice security-wise.

from namecoin-legacy.

You can use multi-threading and multiple processes in a native application with plugin interfaces, too.
Let's take NMControl as an example:
Port 53 -> Python -> random port -> HTTP -> Namecoind-RPC
Script-languages like Python are slow and HTTP has a lot of overhead. Consider the situation of one or two Namecoinds as DNS resolvers of a company/university campus with 5000 hosts. 5000 hosts will overload a Namecoind/NMControl combo.

Namecoin is an excellent theoretical idea but implementations don't seem to consider the scaling effects in the real world ...

from namecoin-legacy.

I understand your concerns. However, I don't see how a plug-in infrastructure would help in the example you cite. The main problem I see with using a namecoind for 5000 clients is that the namecoind itself will be overloaded. It doesn't matter whether it is accessed via RPC or directly from within using plug-ins. At least that is my own impression.

from namecoin-legacy.

I'm with domob on this one. Plugins should go into NMControl. It already does support plugins.

In the example given I don't understand why namecoind would be overloaded. Communication between namecoind and NMControl is not affected by connections to NMControl.

from namecoin-legacy.

I suggest for this discussion to be moved to the forum.

from namecoin-legacy.