Date: Sun, 10 Jul 2011 21:10:07 +0200
From: Pieter Wuille <pieter.wuille@gmail.com>
To: Matt Corallo <bitcoin-list@bluematt.me>
Message-ID: <20110710191005.GA6467@ulyssis.org>
References: <1309801974.3423.80.camel@Desktop666>
	<CABsx9T31ZuQHKwcNnb9-NpaCA6c43PXVZ+Tc+GZ=2Wkz08enHw@mail.gmail.com>
	<1309811972.29355.19.camel@Desktop666>
	<1309828239.29355.28.camel@Desktop666>
	<CABsx9T2-htzpQPxrNqq1Efow2tuUdm5AQGOh0MS0MXvH5pD_5g@mail.gmail.com>
	<CA+8xBpd06DJ_0+OHRDsbzkecfG1+-ubXb2eiVRiSn6GT53ZDSQ@mail.gmail.com>
	<1310307677.2230.5.camel@Desktop666>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1310307677.2230.5.camel@Desktop666>
User-Agent: Mutt/1.5.20 (2009-06-14)
Cc: bitcoin-development <bitcoin-development@lists.sourceforge.net>
Subject: Re: [Bitcoin-development] Encrypted Wallet Backward Compatibility
Precedence: list

On Sun, Jul 10, 2011 at 04:21:17PM +0200, Matt Corallo wrote:
> 
> At Luke's suggestion, I did a bit more digging and was able to find a
> data structure in wallet settings that should cause all versions (well
> all versions since Bitcoin was in github, and probably before then) to
> crash on load instead of making a new wallet or opening in some bizarre
> half-state.  I just put an empty object in addrIncoming (nfc what it was
> used for, but it will get the desire effect and it isnt used anywhere in
> the code aside from its definition).
> You can see the commit at
> https://github.com/TheBlueMatt/bitcoin/commit/2e8383469d7e12a495b3a1dbd41a8d211ff34fe8
> Does anyone disagree and think a different solution would work better?

Though giving an mostly incomprehensible/unrelated error is never nice to
the user, i believe it's better than creating an empty wallet and letting
the user wonder where his wallet went. This way, we fail soon and don't
ever get a corrupt wallet.

> This resolves all known issues and suggestions that I know of on newenc
> except for the invalid mlock calculations, which I will go fix right
> now.  So...aside from that bug does anyone have any remaining
> suggestions/blockers on newenc and, if not, can we get final ACKs on it?

ACK on newenc, and thanks for all the work you put in it already.

-- 
Pieter