Gridcoin Developer Update April 23rd, 2018

Hello folks and welcome to another Developer Update from the Gridcoin team. I'd like to remind everyone that these posts will be created every two weeks unless a wallet update is pending that week.

These last two weeks, developers have been preparing for an upcoming leisure update and finalizing fixes currently in staging branch. Testnet has been busy as well testing various new features and verifying fixes. Some of the pull requests these last two weeks have included:

• #1054 by @thecharlatan cleans up the voting code and removes a call to RPC from within the GUI wallet that has gone unnoticed for a long time.
• #1060 by @tomasbrod is the start of implementing superblock contract forwarding to non-Windows nodes. This is intended as a temporary way to allow non-Windows wallets to help stake superblocks while the current NN is being rewritten to C++. To summarize how this works: when Windows nodes have agreed upon a consensus neural hash, Linux nodes will be able to request the packed superblock contract data associated with the current quorum hash (best hash) and try to stake a superblock just like the Windows nodes. In this implementation the Windows nodes are still doing most of the work gathering the NN data and forming consensus, but the Linux nodes are able to assist in the actual staking of the superblock itself once consensus is reached. This should reduce superblock delays once implemented.
• #1056 by @ifoggz fixes the issue #1035 first reported about a month ago related to the output of beaconreport cutting off the first character of each CPID in the list.
• #1057 by @tomasbrod adds feedback to the addnode command in debug letting the user know whether they successfully connected to the peer or if the connection failed for some reason. This addresses issue #867 and is a nice quality of life improvement.
• #1059 by @ifoggz addresses an issue initially brought to us by the Flyp.me exchange. The relevant issue is #1055. To summarize, the output of the listtransactions RPC command was inconsistent with what was expected from other Bitcoin-like wallets and this PR changes the output to make the call more in line with what exchanges expect to see.

In addition to the PRs listed above, @ifoggz has been working with the testing team to improve performance and better handle errors in the current Windows neural network implementation. So far these tests have been positive and show greatly improved consensus among neural nodes. In particular an issue was found when the downloaded gz files from project sites were incorrectly formatted. In the past, the NN would simply accept the bad data and try to use it for calculations resulting in wildly inaccurate magnitudes and hashes. The new improved NN code will recognize badly formatted project data files and prevent their bad data from being included in NN calculations. The new NN code will also try to redownload the excluded bad data on the next sync.

I would like to thank @Noah-Blaker for all his help assisting in this testing.

I would like to add a quick update on testnet. I gave a "Introduction to Testnet" talk on the last Fireside podcast. Thanks to @jringo for having me and I look forward to hearing the edited audio once it is finalized. The goal of this talk was to educate people about testnet and bring new testers on board.

Thanks for reading this edition of the Developer Update. Expect to see another update two weeks from today (5/7), barring a wallet update around that time. If you have any comments or questions for the Gridcoin development team feel free to ask in the comments below. If I am not able to answer your question directly, I can certainly forward it to someone who can! Thanks everyone!