This is only a bug in dev, but was committed yesterday.
Sometimes the `encrypt` method was being passed values other than the password as the encryption key, leading to un-unlockable vaults.
To find this, and avoid it for all time hereafter, I added several more steps to our oft-neglected integration test suite, which now fully initializes a vault, locks it, and unlocks it again, to make sure all of those steps definitely work always.
If a nodified method does not return a Promise, it will throw an error, like this:
```
Error in event handler for (unknown): Error: The function setSelectedAccount did not return a Promise, but was nodeified.
```
Mostly Fixes#893
A couple methods cache callbacks, and will require a larger refactor to fully denodeify.
Specifically, our methods involving web3 requests to sign a tx, sign a message, and approve or cancel either of those.
I think we should postpone those until the TxManager refactor, since it will likely handle this response caching itself.
At least, the portion of it that we use.
Moved salting within the encryptor, so it does not need to be managed externally.
KeyringController now caches the password instead of a passwordDerivedKey, since it is ignorant of the salt.
Encryptor payload is now in a JSON format, so its portions are both base64 encoded *and* labeled appropriately. The format is `{ "data": "0x0", "iv": "0x0", "salt": "string" }`.
Fixes#843Fixes#859
Ropsten links will still not work until Etherscan publishes their ropsten link format.
At that time we will need to update ui/lib/account-link.js
Otherwise, fixes#831
Fixes#791
It was possible for two requests to have the same ID, causing a crash and loss of StreamProvider connection.
This new id generation strategy creates a random ID, and increments it for each request.
In case the id generator is included from two different processes, I'm initializing the counter at a random number, and rolling it over a large number when it gets too big.
Our gas price buffering logic had a bug, because bn.js has inconsistent behavior when using hex-prefixed output. The issue has been opened with them here:
https://github.com/indutny/bn.js/issues/151
We've corrected our usage in the mean time.
Now old vaults are recognized as an "Initialized" MetaMask instance.
Upon logging in, when fetching the initial password-derived key, if there is no new-style vault, but there is an old style vault, it is migrated to the new format before proceeding through the usual unlocking steps.
This contract hex does include the value `f4`, but it was compiled from a contract with no instance of `.delegatecall`. I believe `f4` in this case is part of some other value or contract address, and `ethBinToOps` has some error in how it skips pushed data.
@kumavis
For some reason the popup was often cutting off the bottom buttons of the UI.
We should look at that more carefully later perhaps, but especially since we're considering moving off the popup, I'm just fixing it by making it taller for now.
SubmitPassword was not creating a new id-management
This is because I broke up the old "createIdmgmt" method to not perform as much conditional logic.
Now the pieces are reusable and do what they should do.
Fixed bug where the second new vault created in an IdStore would initially return the accounts from the original store.
Also fixed some tests that were incorrect.
eth-lightwallet was previously not salting vault passwords, potentially making it easier to crack them once obtained.
This branch incorporates the API changes to allow us to take advantage of the new salting logic.
This is still throwing deprecation warnings, but that's actually a bug in eth-lightwallet I wrote, [I've submitted a PR for that here](https://github.com/ConsenSys/eth-lightwallet/pull/116).
Fixes#555
Previously the metamask controller only supported a single UI event listener, which wasn't useful for having a separate notification UI open at the same time.
Also reduced the notification's complexity down to a single method, which is heavily re-used.
Still has an outstanding bug where if the plugin ui dismisses the last tx, it does not close the notification popup.
I'm unsure which will be more performant: A notification using a trimmed down version of the UI, or using them both, letting the browser cache them both.
In any case, here I've modified the normal UI to recognize when it's a popup, and change the UX accordingly in a few ways:
- Hide the menu bar
- Hide the back button from the notifications view.
- When confirming the last tx, close the window.
* Add platform specific folders to dist folder
* Remove gulp hacks
* Add platform specific bundling
dev and dist tasks now build into platform-specific folders within the `dist` folder.
Added tasks `gulp zip` and `gulp dist`.
`zip` builds the platform-specific folders into platform-specific bundles within the `dist` folder.
`dist` builds and then zips all at once.
* Fix chrome bundle zipping
* Fix broken reference in eth warning
* Fix but where web3.eth.accounts are not available in firefox.
* Bump changelog
* WIP: Buy button link
* Add buy eth and the buy eth warning message
* Add css
* Move the opening of coinbase page to background
and send to faucet if on test net
* Create a Warning about storeing eth
* Finish Buy button and Eth store warning screen
* Add to CHANGELOG
* fix frankies deletion and change chrome to extension
* Add mozilla plugin key to manifest
* Move all chrome references into platform-checking module
Addresses #453
* Add chrome global back to linter blacklist
* Add tests
Replace config-manager-singleton with one that is instantiated within the `background-controller`, and takes its persistence callbacks from its instantiated options.
Replaced class getters with more familiar get___() functions.
Signing now always takes a 64 digit hex string, and returns a message signature which appropriately pads r, s, and v with zeroes.
Need to verify with Denis that this is the behavior he requires.
Scrolling to the bottom of the accounts page now reveals a downward-facing chevron button.
Pressing this button shows loading indication, adds a new account to the identity vault, displays it in the list, and scrolls the list to the bottom of the page.
Any number of accounts can be generated in this way, and the UX feels intuitive without having to overly explain how HD paths work.
Fixes#122
Had used multiple actions for some transitions, which would lead to brief intermediary states.
Now making a few actions much more explicit about what they route to, so there is less intermediary logic, and we can transition confidently to the correct view.
No longer do our `mainnet` and `testnet` buttons set specific RPC urls. Now they set `provider.type`, which gets interpreted with code.
Currently the provider types of `mainnet` and `testnet` point to our new scalable backends, but these could be re-interpreted to use any other provider, be it etherscan, peer to peer, or otherwise.
Makes it easier for us to upgrade our infrastructure without incorporating migration logic into the program.
Calls to `eth.sign` are now transiently persisted in memory, and displayed in a chronological stack with pending transactions (which are still persisted to disk).
This allows the user a method to sign/cancel transactions even if they miss the Chrome notification.
Improved a lot of the view routing, to avoid cases where routes would show an empty account view, or transition to the accounts list when it shouldn't.
Broke the transaction approval view into a couple components so messages and transactions could have their own templates.
When starting up, we now create a `web3` inside the `background.js` process, which we pass to the `idStore` and ask for the current `network`.
We include the `network` on `app.metamask.network` in the state object.
We re-request the network when changing provider.
We filter the transaction list for transactions that match the current network.
When selecting an account, we now persist the selection to the `configManager`, so the selection can be restored when re-unlocking Metamask.
Also found the bug where `rawtestrpc` was still being used as a default, and fixed it!
- When unlocking, the first account is now selected by default and displayed as the main view.
- There is now a "CHANGE ACCT" button on the detail view to show the accounts list.
- Clicking an account from the accounts list now navigates to the detail view and selects that account.
- Config/Info screen "back" buttons now fire a new action, `GO_HOME`, which is configured to navigate to the accountDetail view, putting that logic in one place.
- When locking and unlocking again, the first account is always displayed, eventually we should persist the selection.
Transactions are now stored, and are never deleted, they only have their status updated.
We can add deleting later if we'd like.
I've hacked on emitting the new unconfirmedTx key to the UI in the format it received before, I want Aaron's opinion on where I should actually do that.
This boolean is computed from these requirements:
- The user is on the testnet rpc
- The account is index 0
The UI is responsible for checking the account balancing and indicating if fauceting is indeed pending or not.
Abstract all configuration data into a singleton called `configManager`, who is responsible for reading and writing to the persisted storage (localStorage, in our case).
Uses my new module [pojo-migrator](https://www.npmjs.com/package/pojo-migrator), and wraps it with the `ConfigManager` class, which we can hang any state setting or getting methods we need.
By keeping all the persisted state in one place, we can stabilize its outward-facing API, making the interactions increasingly atomic, which will allow us to add features that require restructuring the persisted data in the long term without having to rewrite UI or even `background.js` code.
All the restructuring and data-type management is kept in one neat little place.
This should make it very easy to add new configuration options like user-configured providers, per-domain vaults, and more!
I know this doesn't seem like a big user-facing feature, but we have a big laundry list of features that I think this will really help streamline.
Also added the hdPath that Christian had told me to our calls to the LightWallet, but this does not seem to have made us generate the same accounts as `testrpc` yet.
Added initial test just to verify we can recover the accounts we generate in this way.
Still need to add compliance test to make sure this interoperates with testrpc's new mnemonic flag.
In order to persist the seed word page until the user clicks the confirmation button, we need to store the seed words in localStorage.
To simplify this process I've also reorganized some of the account manager code, broken up one large function into many smaller functions, and created a new class for the IdMgmt object.
Again, sorry such a big refactor in one commit, but I really had to break it down to work through it.
