Adds support for the 'get_finished' and 'get_peer_finished' functions

[movitto]

Returns latest 'finished' messages sent and received.

[ioquatix]

Can you rebase on master and explain why these functions are useful?

[ioquatix]

@movitto maybe add a simple example to the documentation, would be sufficient to explain how to use it and why.

[movitto]

@ioquatix thanks for taking a look at this. I just rebased and force pushed, the branch should now be up to date.

We need this to incorporate it in a ruby implementation of the rippled (server behind the XRP blockchain/crypto-currency) peer-communication protocol. During the initial handshake, rippled uses the get_finished and get_peer_finished functions as the basis of a unique "session signature" used to authenticate the session. I can't attest to the design considerations that went into using these functions in that capacity (I don't work for ripple labs, nor was I involved in those discussions) but for the time being we'll need access to it from ruby to complete the handshake.

You can see how we use it in our XRP client library.

Thanks again for looking at this and feel free to let me know if there is anything else I can do to get this in!

[ioquatix]

I think there is one issue - this won't compile on Windows because even thought we have C99, we don't have support for VLA on Windows compiler (wtf Microsoft).

char buf[len+1];

Can probably be replaced with

char * buffer = ALLOCA_N(char, len+1);

@mame do you mind commenting if this is the right approach?

[ioquatix]

Rename this to finished_message

[ioquatix]

Use ALLOCA_N as suggested.

[ioquatix]

Rename this to peer_finished_message

[ioquatix]

Sorry to make this more complicated for what is a fairly simple feature, but some tests would be great too.

I mentioned in the review should we call this socket.finished or socket.finished_message. To me, it's a bit confusing because finished isn't a noun. That's the only reason why I suggested finished_message.

[movitto]

@ioquatix ok cool, I addressed all feedback:

• 'peer_finished' is now known as 'finished_message'
• 'get_peer_finished' is now known as 'peer_finished_message'
• VLAs have been replaced with ALLOCA_N(char, len+1);
• test has been included verifying equality of client/server finished_message / peer_finished_message (the 'start_server' test helper had to be extended to accept a callback so that we could access server socket)
• rebased against master and tidy'd up the patchset

Let me know if there is anything else!

[ioquatix]

If you can rebase this on master and all tests pass I will merge.

[ioquatix]

Thanks for your effort!

[movitto]

@ioquatix thanks for merging!