Re: [Bitcoin-development] Payment Protocol for Face-to-face Payments

[Mike Hearn <mike@plan99.net>]

[-- Attachment #1: Type: text/plain, Size: 1607 bytes --]

On Fri, Mar 21, 2014 at 11:59 AM, Adam Back <adam@cypherspace.org> wrote:

> Maybe its time to explore raw ECDSA signed message based certs.
>

If you want to create and run a new CA, by all means. But I bet you don't.
So we're stuck with the current system for now.


> btw I dont think its quite 4kB.  eg bitpay's looks to be about 1.5kB in der
> format.  And they contain a 2048-bit RSA server key, and 2048-bit RSA
> signatures (256byte each right there = 512bytes).  And even 2048 is weaker
> than 256-bit ECDSA.


But you have to chain up to the root.

The only reason more certs aren't ECC is backwards compatibility. Some old
browsers don't know how to handle them. It wasn't so long ago that Fedora
and Android were deleting ECC code from upstream libraries before shipping
them, either for patent reasons for disk space saving measures.

But it's possible to get ECC certs if you want. For example, Entrust is
starting to sell them:

http://www.entrust.net/ecc-certs/index.htm

But their intermediate cert is still RSA. My understanding is that ECC
roots for many CA's have been submitted and are now included, but of course
"give up compatibility with lots of users" vs "save a bit of cpu time and a
handful of bytes" is no real competition so it will be a long time until
most websites are using ECC certs.

Regardless, it's all irrelevant. Who knows when we might want to add
another feature that uses some bytes into PaymentRequests. Stuffing them
into a QR code will never make much sense IMO - it's far more sensible to
just use Bluetooth where the data size constraints are so much easier.

[-- Attachment #2: Type: text/html, Size: 2454 bytes --]