BIP-0062 Canonical Signatures #2500
Conversation
|
Looks reasonable to me. |
mvandeberg
force-pushed
the
1944-enforce-canonical-signatures
branch
from
b90f5b6 to
414ac3f
Compare
mvandeberg
force-pushed
the
1944-enforce-canonical-signatures
branch
from
9af0755 to
63fc2a8
Compare
goldibex
left a comment
goldibex
left a comment
There was a problem hiding this comment.
In general -- reading this PR makes me realize we need a way to test transitions across hardforks. It would be really nice to have a test that
- tries to submit a transaction with a non-FC-canonical but BIP-0062-canonical signature pre-fork and asserts failure
- joins HF20
- submits the transaction again and asserts success
| const authority_getter& get_owner, | ||
| const authority_getter& get_posting, | ||
| uint32_t max_recursion = STEEM_MAX_SIG_CHECK_DEPTH | ||
| uint32_t max_recursion/* = STEEM_MAX_SIG_CHECK_DEPTH*/, |
There was a problem hiding this comment.
Please just delete these defaults rather than comment them out
There was a problem hiding this comment.
I left the defaults there because there is some cleanup that needs occur post HF that will cleanup this interface and restore it to its pre hardfork state.
| for( size_t i = 0; i < 4; i++ ) | ||
| { | ||
| sig <<= 64; | ||
| sig += boost::endian::big_to_native( *( uint64_t* )( c.data + 33 + ( i * 8 ) ) ); |
There was a problem hiding this comment.
I'm not convinced all this complexity is necessary. Since the numbers are both guaranteed to be big-endian and fixed-length, a simple memcmp should be all that's necessary to determine which is the larger number.
goldibex
left a comment
goldibex
left a comment
There was a problem hiding this comment.
See previous comments. memcmp or the equivalent c++ sugar is the easiest way to verify the supplied signature is smaller than N/2.
|
lgtm |
Closes #1944