Update Frontier #307
Update Frontier #307
Conversation
pirapira
force-pushed
the
frontier
branch
2 times, most recently
from
5b493cf
to
47a8195
Compare
|
Maybe this type of "backporting" PR will be opened many times (for Frontier and others). Do you have a system in mind to track this? |
|
Note that the merge conflicts need to be fixed and there are other changes that have been made and are pending in PRs that would apply to previous releases. PR titles and commit summaries should include an abbreviate to indicate whether they are a release-specific change or not, in order to help systematize these kind of changes. E.g. RS = Release-specific. GC = general change (not release specific), applies to all releases. Either should be explicitly included. |
|
Or better than or in addition to RS: have the specific release. Changes in a release may apply in future releases, unless overwritten or major protocol-wide sweeping changes occur, e.g. abstraction, Casper, statelessness, etc. |
… is already used nearby for the set of nibble sequences.
This commit adds notes to MLOAD, MSTORE and MSTORE8 specification
saying the modulo is not considered when computing the new memory
size.
Before this change, since the beginning of the table contains
`All arithmetic is modulo $2^{256}$ unless otherwise noted.`,
the memory size computation was affected by the modulo computation too.
For instance, MLOAD with the biggest uint256 didn't change the memory size
most of the time.
Before this commit, equation (100) was not satisfiable when s and r were equal and v + v was not zero. Though this equation is further complemented in equations (101) through (104), I thought (100) needed a condition s != r.
This solves ethereum#190. The other way of putting parentheses did not make sense becasue `sign(x) * |x|` can immediately be optimized into `x`. Also, I checked one EVM implementation's behavior ``` evm --debug --code "60601960401907" ``` and saw SMOD returning a negative-looking number.
…ng gas from the machine state
as in other occurrences.
Also cleanup nomenclature around byte sequences.
…catenate a quadruple with an element
This PR updates the frontier branch, cherry-picking the commits in
masterthat are not about the more recent hardforks.