Mike Hearn [ARCHIVE] on Nostr: 📅 Original date posted:2014-03-06 📝 Original message:On Thu, Mar 6, 2014 at ...
📅 Original date posted:2014-03-06
📝 Original message:On Thu, Mar 6, 2014 at 12:26 PM, Andreas Schildbach
<andreas at schildbach.de>wrote:
> I'm not sure if iso-dep is the way to go here. Afaik as soon as you pick
> up the phone the connection breaks.
If the phone isn't willing to immediately authorise then it'd have to fall
back to HTTPS or Bluetooth as normal.
> Besides, how do you plan to risk-analyse the memo field?
>
I guess only the amount and destination are relevant for risk analysis.
> It's already very short if you can do without Android Beam, e.g. on
> Android 2.3.
I think IsoDep based protocols must bypass Beam - when I scan my e-passport
there's no beam animation.
> The most obvious optimization to speed up signature checking is to make
> it lazy. The user can already inspect the payment while signatures are
> being checked.
Well, for <400msec there can't be any user interaction. But checking
signatures on the payment request and constructing and signing the inputs
can all be done in parallel - you should be able to max out every core, at
least for a brief moment.
> Even the current ~10 second roundtrip is a huge improvement to the
> status quo. I recently tried to buy a subway ticket and it took me 7
> full minutes (just for the payment process)!
Then that subway kind of sucks ;) Have you been to London and used Oyster?
I think the capital wouldn't work at all without the low latency Oyster
cards. The tube would have stopped scaling some time ago.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxfoundation.org/pipermail/bitcoin-dev/attachments/20140306/f3881a64/attachment.html>
📝 Original message:On Thu, Mar 6, 2014 at 12:26 PM, Andreas Schildbach
<andreas at schildbach.de>wrote:
> I'm not sure if iso-dep is the way to go here. Afaik as soon as you pick
> up the phone the connection breaks.
If the phone isn't willing to immediately authorise then it'd have to fall
back to HTTPS or Bluetooth as normal.
> Besides, how do you plan to risk-analyse the memo field?
>
I guess only the amount and destination are relevant for risk analysis.
> It's already very short if you can do without Android Beam, e.g. on
> Android 2.3.
I think IsoDep based protocols must bypass Beam - when I scan my e-passport
there's no beam animation.
> The most obvious optimization to speed up signature checking is to make
> it lazy. The user can already inspect the payment while signatures are
> being checked.
Well, for <400msec there can't be any user interaction. But checking
signatures on the payment request and constructing and signing the inputs
can all be done in parallel - you should be able to max out every core, at
least for a brief moment.
> Even the current ~10 second roundtrip is a huge improvement to the
> status quo. I recently tried to buy a subway ticket and it took me 7
> full minutes (just for the payment process)!
Then that subway kind of sucks ;) Have you been to London and used Oyster?
I think the capital wouldn't work at all without the low latency Oyster
cards. The tube would have stopped scaling some time ago.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxfoundation.org/pipermail/bitcoin-dev/attachments/20140306/f3881a64/attachment.html>