Low min difficulty, and smoother adjustment

[dooglus]

This is a hard-fork change.

It will allow us to have 1-minute blocks, rather than the 3 minute blocks we're currently seeing.

It lowers the minimum staking difficulty to 16 times lower, and uses the average time per block of the last 16 blocks rather than just the last 2, so difficulty won't jump around as much.

This will need editing (in main.cpp):

const int LOWER_DIFFICULTY_START = 42000;

based on when we want the hard fork to happen.

I've not been able to get a testnet instance running, because I'm stuck on the PoW stage and so this code isn't very well tested yet.

Are you able to test this change?

[l0rdicon]

Yes I can. I'll also document the testnet setup for us and get a base chain we can use in the future with with the pow and reasonable send list completed.

I'l a little under the weather but this is high on my to do list, will have this ready tonight or tomorrow for sure.

This can be merged into develope any time, when I get to testing I'll merge it in if it hasn't already.

I'm also going to create the staging branch I referred to. The idea would be develope->staging->master and small fixes updates can go directly into staging before being merged into master.

I believe thats a reasonable way to organize things but suggestions are also welcome

[dooglus]

I read http://nvie.com/posts/a-successful-git-branching-model/ and agree that it's a great way to use git.

It's pretty much exactly the same model I've used in the past (except we never deleted branches when we were done with them, so things got incredibly messy).

I'll merge this right now.

[dooglus]

I figured I can test this as follows:

1. back up ~/.clam
2. add "connect=127.0.0.1" to clam.conf
3. change the LOWER_DIFFICULTY_START value to the current block
4. run clam-qt as usual
5. watch what happens
6. stop and revert to backup

That way I can use mainnet as if it was testnet without affecting anything.

I'll post my findings.

[dooglus]

http://pastie.org/pastes/9414363/text is a log of how the new difficulty adjustment testing went

I had a few interruptions, causing it to look like some blocks were very slow.

The difficulty was adjusted to get the average speed back on track, as expected, but using a moving average maybe causes 'overshoots'.