Received-SPF: pass (sog-mx-2.v43.ch3.sourceforge.com: domain of taplink.co
	designates 50.117.27.232 as permitted sender)
	client-ip=50.117.27.232; envelope-from=jeremy@taplink.co;
	helo=mail.taplink.co; 
Content-Type: multipart/alternative; boundary=----------svoj35mbC6KPQXlw6JqQzt
To: "bitcoin-development@lists.sourceforge.net"
	<bitcoin-development@lists.sourceforge.net>
References: <20140106120338.GA14918@savin>
	<op.w9c5o7vgyldrnw@laptop-air.hsd1.ca.comcast.net>
	<20140110102037.GB25749@savin>
	<op.w9kkxcityldrnw@laptop-air.hsd1.ca.comcast.net>
	<CANEZrP0Np_FGhw=m6OffijByzz9r4D2AA78jCkzh=NZh=xrbjQ@mail.gmail.com>
Date: Sun, 12 Jan 2014 10:20:18 -0800
MIME-Version: 1.0
From: "Jeremy Spilman" <jeremy@taplink.co>
Organization: TapLink
Message-ID: <op.w9k6j4xryldrnw@laptop-air.hsd1.ca.comcast.net>
In-Reply-To: <CANEZrP0Np_FGhw=m6OffijByzz9r4D2AA78jCkzh=NZh=xrbjQ@mail.gmail.com>
User-Agent: Opera Mail/1.0 (Win32)
oclient: 192.168.168.135#jeremy@taplink.co#465
Subject: Re: [Bitcoin-development] Stealth Addresses
Precedence: list

------------svoj35mbC6KPQXlw6JqQzt
Content-Type: text/plain; charset=iso-8859-15; format=flowed; delsp=yes
Content-Transfer-Encoding: 7bit


> You can always just extend the payment protocol with the new fields as  
> well, vs making very long addresses.

I should have mentioned that as Task #4. Agree it could be an optional  
extension and backward compatible. I think for displaying the payment in  
the UI after it's been made via PP, we have to fully support sending to a  
new standard address type anyway. Probably easiest to implement in PP  
after the address and transaction building code is done.

So '4a' would be building a static PP file given the necessary inputs.  
When I get to that point, I'll send out a draft PP extension with  
fields/formats if someone else hasn't already. '4b' would be actually  
adding support for parsing those fields and generating the new transaction  
type into bitcoind.

Any thoughts on the prefix, and one vs two pubkey approach? First of all,  
do we try to support both equally, or favor one over the other? I was  
thinking we could have two different 4 byte prefixes but that both render  
as xSTL/tSTL in Base58 but correspond to the one vs two pubkeys expected.  
I think the chance of finding a single prefix which looks like xSTL for  
both address lengths is 1 in (58^4)^2, so that's probably not going to  
happen.

 From the payer/user perspective, short stealth vs. long stealth is  
irrelevant; they both have the same usability properties from the payer  
perspective. So giving them the same Base58 prefix seems like a good plan.

The full 4-byte prefix seems worth the usability trade-off versus 1-byte  
prefix, especially since it will impact the ability to lookup the  
transaction on an outside service, which I think a lot of people do to  
verify their payments. IMO a longer prefix isn't "wasting bytes" anywhere  
that it really counts.

We could save two bytes in the address if we required both pubkeys to  
start with '03', or save one byte if we required they both start with the  
same byte, but again doesn't seem worth it (to me) for the arbitrary  
restriction.

The actual internal wallet code for *receiving* STL payments and updating  
balances is more tricky and probably not something I can personally tackle  
for bitcoind. Assuming we even want first-class support for generating STL  
addresses and receiving STL payments in a standard user wallet, someone  
has to decide if the STL 'd' / 'd2' keys should be...

   1) Encrypted as usual, and then keep a list of blocks with interesting  
transactions, and go through them when the user enters their password?   
This would cause balances to update differently than how they do now, but  
perhaps be more secure.

   2) Kept unencrypted to allow live scanning as usual? Or keep just 'd2'  
unencrypted, with some new concept of 'unconfirmed' until the user enters  
their password to prove they can spend that TX? That kind of extra step  
seems OK for a merchant but sounds very scary for an average user.

   3) Kept encrypted under a separate password? Meh...

And last thought for now... At some point, we might want to decide on a  
convention to highlight these STL addresses as 'reusable' -- but similar  
questions around revocability remain. I hope we don't need anything like a  
UTC expiration time baked in to the address. A static PP file will have an  
expiration date either in the certificate or in 'expires' field, so I  
think if you want it to expire then use PP?
------------svoj35mbC6KPQXlw6JqQzt
Content-Type: multipart/related; boundary=----------svoj35mbC6KPQX57ksPHtp

------------svoj35mbC6KPQX57ksPHtp
Content-Type: text/html; charset=iso-8859-15
Content-ID: <op.1389550818119.875f0a785d6b323e@192.168.168.135>
Content-Transfer-Encoding: Quoted-Printable

<!DOCTYPE html><html><head>
<style type=3D"text/css">body { font-family:'Times New Roman'; font-size=
:13px}</style>
</head>
<body><br><blockquote style=3D"margin: 0 0 0.80ex; border-left: #0000FF =
2px solid; padding-left: 1ex"><div dir=3D"ltr">You can always just exten=
d the payment protocol with the new fields as well, vs making very long =
addresses.&nbsp;</div></blockquote><div><br></div><div>I should have men=
tioned that as Task #4. Agree it could be an optional extension and back=
ward compatible. I think for displaying the payment in the UI after it's=
 been made via PP, we have to fully support sending to a new standard ad=
dress type anyway. Probably easiest to implement in PP after the address=
 and transaction building code is done.</div><div><br></div><div>So '4a'=
 would be building a static PP file given the necessary inputs. When I g=
et to that point, I'll send out a draft PP extension with fields/formats=
 if someone else hasn't already. '4b' would be actually adding support f=
or parsing those fields and generating the new transaction type into bit=
coind.</div><div><br></div><div>Any thoughts on the prefix, and one vs t=
wo pubkey approach? First of all, do we try to support both equally, or =
favor one over the other? I was thinking we could have two different 4 b=
yte prefixes but that both render as xSTL/tSTL in Base58 but correspond =
to the one vs two pubkeys expected. I think the chance of finding a sing=
le prefix which looks like xSTL for both address lengths is 1 in (58^4)^=
2, so that's probably not going to happen.</div><div><br></div><div>From=
 the payer/user perspective, short stealth vs. long stealth is irrelevan=
t; they both have the same usability properties from the payer perspecti=
ve. So giving them the same Base58 prefix seems like a good plan.</div><=
div><br></div><div>The full 4-byte prefix seems worth the usability trad=
e-off versus 1-byte prefix, especially since it will impact the ability =
to lookup the transaction on an outside service, which I think a lot of =
people do to verify their payments. IMO a longer prefix isn't "wasting b=
ytes" anywhere that it really counts.</div><div><br></div><div>We could =
save two bytes in the address if we required both pubkeys to start with =
'03', or save one byte if we required they both start with the same byte=
, but again doesn't seem worth it (to me) for the arbitrary restriction.=
</div><div><br></div><div>The actual internal wallet code for *receiving=
* STL payments and updating balances is more tricky and probably not som=
ething I can personally tackle for bitcoind. Assuming we even want first=
-class support for generating STL addresses and receiving STL payments i=
n a standard user wallet, someone has to decide if the STL 'd' / 'd2' ke=
ys should be...</div><div><br></div><div>&nbsp; 1) Encrypted as usual, a=
nd then keep a list of blocks with interesting transactions, and go thro=
ugh them when the user enters their password? &nbsp;This would cause bal=
ances to update differently than how they do now, but perhaps be more se=
cure.</div><div><br></div><div>&nbsp; 2) Kept unencrypted to allow live =
scanning as usual? Or keep just 'd2' unencrypted, with some new concept =
of 'unconfirmed' until the user enters their password to prove they can =
spend that TX? That kind of extra step seems OK for a merchant but sound=
s very scary for an average user.</div><div><br></div><div>&nbsp; 3) Kep=
t encrypted under a separate password? Meh...</div><div><br></div><div>A=
nd last thought for now... At some point, we might want to decide on a c=
onvention to highlight these STL addresses as 'reusable' -- but similar =
questions around revocability remain. I hope we don't need anything like=
 a UTC expiration time baked in to the address. A static PP file will ha=
ve an expiration date either in the certificate or in 'expires' field, s=
o I think if you want it to expire then use PP?</div></body></html>
------------svoj35mbC6KPQX57ksPHtp--

------------svoj35mbC6KPQXlw6JqQzt--