Accounting and the reference client

Ross Nicoll bio photo By Ross Nicoll

Bit of an odd update this week, but as the issue is currently highly visible, it seems a good time to say this:

Do not use the accounting system in the reference client for anything important.

If you're not using RPC (the commands listed under https://en.bitcoin.it/w/index.php?title=Original_Bitcoin_client/API_calls_list&oldid=48664 ), this probably won't mean much to you, and you can probably safely ignore this post.

While I'm not aware of the full details of the emergency migration of Moolah's platform from V1 to V2, I do know that the accounting system in the reference client was part of the problem with V1. This isn't a Dogecoin-specific matter, this is something we've inherited from Bitcoin Core. The accounting system has a lot of limitations, and the opinion from the Bitcoin Core developers is that it's better to simply remove it ( https://github.com/bitcoin/bitcoin/issues/3816 ) than trying to fix it.

There's no timescale for removing the functionality yet, but you can expect that in time it will be cut out of the client. If you're already using the accounting system for an existing service, please start working on extracting transaction data to your own database, and developing an accounting system from there.

If you're considering writing a new service, please do not attempt to use the integrated accounting system.

If anyone wants to write a framework for extracting transaction data regularly and synchronising it into a database, and running accounting from that database, I believe many would be incredibly grateful.

Beyond that; a 1.7.2 client beta should be out essentially as soon as we get enough developers into a room together to agree it's ready. This brings in a huge number of bug fixes, and will be highly recommended (but still not required) for all users currently on the 1.6 client.

Lastly, we're starting to see UI mockups showing up; the development team are focusing on keeping the reference client minimal, and personally I hope to reduce the interaction with the wallet, not add more to it. Payment protocol will help with this, and ideally I'm hoping we can develop a protocol to allow sites to request addresses from the wallet automatically (without a cut & paste step). Less is definitely more here, especially as any added features increase the amount of code in which subtle security issues can potentially arise.

Others are of course welcome to fork the reference client and extend it. My preferred option would be to consider a stand-alone "Social dashboard" application, if there's definitely a desire to do this, although most of the options listed would work just as well from an HTML 5 website IMHO.

Next couple of weeks are going to be crazy busy my end, so don't panic if you don't hear from me before August. I'll be around, as will the other devs, just perhaps a little quieter.