* adding mock contacts, test code
* adding changes to ERC721.sol per @frangio's comments on original PR #1943
* fix solhint warnings
* Update contracts/token/ERC721/ERC721.sol
Co-Authored-By: Francisco Giordano <frangio.1@gmail.com>
* same revert wording per @frangio's review suggestion
* per @frangio's feedback, changing the inline assembly to accomplish: we want to ignore the first 4 bytes of content, so we should read the length and decrease it by 4, then take the memory location and add 4 to it, then store the new length at the new memory location, then that is the new byte array that we want.
* change revert msg assembly per PR comment by @frangio
* unify revert msg in test code
* fix some failed tests, wording change
* Update contracts/token/ERC721/ERC721.sol
Co-Authored-By: Francisco Giordano <frangio.1@gmail.com>
* Update contracts/token/ERC721/ERC721.sol
Co-Authored-By: Francisco Giordano <frangio.1@gmail.com>
* fix test case, revert without reason
* fix 'ERC721ReceiverRevertsMock: Transaction rejected by receiver'
* style change per review by @frangio
* fix revert reason forwarding
* remove duplicate contracts/mocks/ERC721ReceiverRevertsMock.sol per review https://github.com/OpenZeppelin/openzeppelin-contracts/pull/2018\#issuecomment-574381034
* Add changelog entry
* Fix tests
* Make tests more clear
Co-authored-by: Francisco Giordano <frangio.1@gmail.com>
Co-authored-by: Nicolás Venturo <nicolas.venturo@gmail.com>
* Update ReentrancyGuard for Istanbul Hard Fork
Changes:
Added L37, `_guardCounter = 1;`
Rationale:
The planned _Istanbul Hard Fork_ will implement [EIP 2200](e4d4ea348e/EIPS/eip-2200.md), which implements "net gas metering" for `sstore` operations. If the final value of `_guardCounter` is unchanged relative to the original value of it, a gas refund will be applied and charges for changing the value of `_guardCounter` will effectively not exist. This ends up being cheaper than the current implementation ONLY AFTER Istanbul. Before Istanbul, the added line actually ends up costing more gas.
Note that if `_guardCounter` is `0` initially, the initial cost and subsequent refund will both be larger than if `_guardCounter` is `1` initially. Although in both cases, the net gas cost (`gasCost - gasRefund`) are equal, it's better in terms of cost to have both the gas cost and refund smaller, as there is some limit to the percentage of a gas refund that can actually be realized.
* Update CHANGELOG.md
Added note for change to ReentrancyGuard.sol
* Update ReentrancyGuard.sol
* Update CHANGELOG.md
* Add base Context contract
* Add GSNContext and tests
* Add RelayHub deployment to tests
* Add RelayProvider integration, complete GSNContext tests
* Switch dependency to openzeppelin-gsn-provider
* Add default txfee to provider
* Add basic signing recipient
* Sign more values
* Add comment clarifying RelayHub's msg.data
* Make context constructors internal
* Rename SigningRecipient to GSNRecipientSignedData
* Add ERC20Charge recipients
* Harcode RelayHub address into GSNContext
* Fix Solidity linter errors
* Run server from binary, use gsn-helpers to fund it
* Migrate to published @openzeppelin/gsn-helpers
* Silence false-positive compiler warning
* Use GSN helper assertions
* Rename meta-tx to gsn, take out of drafts
* Merge ERC20 charge recipients into a single one
* Rename GSNRecipients to Bouncers
* Add GSNBouncerUtils to decouple the bouncers from GSNRecipient
* Add _upgradeRelayHub
* Store RelayHub address using unstructored storage
* Add IRelayHub
* Add _withdrawDeposits to GSNRecipient
* Add relayHub version to recipient
* Make _acceptRelayedCall and _declineRelayedCall easier to use
* Rename GSNBouncerUtils to GSNBouncerBase, make it IRelayRecipient
* Improve GSNBouncerBase, make pre and post sender-protected and optional
* Fix GSNBouncerERC20Fee, add tests
* Add missing GSNBouncerSignature test
* Override transferFrom in __unstable__ERC20PrimaryAdmin
* Fix gsn dependencies in package.json
* Rhub address slot reduced by 1
* Rename relay hub changed event
* Use released gsn-provider
* Run relayer with short sleep of 1s instead of 100ms
* update package-lock.json
* clear circle cache
* use optimized gsn-provider
* update to latest @openzeppelin/gsn-provider
* replace with gsn dev provider
* remove relay server
* rename arguments in approveFunction
* fix GSNBouncerSignature test
* change gsn txfee
* initialize development provider only once
* update RelayHub interface
* adapt to new IRelayHub.withdraw
* update @openzeppelin/gsn-helpers
* update relayhub singleton address
* fix helper name
* set up gsn provider for coverage too
* lint
* Revert "set up gsn provider for coverage too"
This reverts commit 8a7b5be5f9.
* remove unused code
* add gsn provider to coverage
* move truffle contract options back out
* increase gas limit for coverage
* remove unreachable code
* add more gas for GSNContext test
* fix test suite name
* rename GSNBouncerBase internal API
* remove onlyRelayHub modifier
* add explicit inheritance
* remove redundant event
* update name of bouncers error codes enums
* add basic docs page for gsn contracts
* make gsn directory all caps
* add changelog entry
* lint
* enable test run to fail in coverage
* Using extcodehash instead of extcodesize for less gas
`extcodehash` uses less gas then `extcodesize`. You can tell which address is a contract by the hash (see EIP-1052).
* Fix
* Add explainer
* Update Address.sol
* add changelog entry
* Updated code style to no more than120 characters per line.
* Unify code comments style with Doxygen-style tags.
* Fix the conflicts.
* Add a return value in the contract ERC20Burnable.
* A Add a wrapper function to change type of address to address payable.
* U Modify Address utils.
* A Add test case for Address.
* U Modify code style in ERC20Burnable.
* Add changelog entry.
* Improved dev docs.
* Transaction Malleability:
If you allow for both values 0/1 and 27/28, you allow two different
signatures both resulting in a same valid recovery. (r,s,0/1) and
(r,s,27/28) would both be valid, recover the same public key and sign
the same data. Furthermore, given (r,s,0/1), (r,s,27/28) can be
constructed by anyone.
* Transaction Malleability:
EIP-2 still allows signature malleabality for ecrecover(), remove this
possibility and force the signature to be unique.
* Added a reference to appendix F to the yellow paper and improved
comment.
* better test description for testing the version 0, which returns
a zero address
* Check that the conversion from 0/1 to 27/28 only happens if its 0/1
* improved formatting
* Refactor ECDSA code a bit.
* Refactor ECDSA tests a bit.
* Add changelog entry.
* Add high-s check test.
Hao (Alan) Tang
Francisco Giordano
Nicolás Venturo
Ignacio Mazzara
ericDeCourcy
Benjamin Smith
Aram Elchyan
Philippe Castonguay
skyge
Bertrand Masius
Thomas Bocek