Backport #8200 to v7.7.9. Original commit description:
The method registry was being initialized with the global variable
`ethereumProvider` before that variable was set. As a result, the
method registry was falling back to an internally constructed provider
that used the wrong provider URL (an obsolete Infura API). This was
resulting in an error with the message "Project ID not found".
The method registry is now initialized lazily, when it's first needed.
This should be well after the initialization of `ethereumProvider`,
which occurs during the UI initialization.
Backport #8363 to v7.7.9. Note that this uses `clone` instead of
`cloneDeep`, because `clone` hadn't yet been replaced by `cloneDeep` on
`master`.
Backporting that change as well would have been very disruptive, so
I've updated this to use `clone` instead to minimize conflicts. It is
functionally equivalent.
Co-authored-by: Whymarrh Whitby <whymarrh.whitby@gmail.com>
This backports the deletion of the Dai/Sai migration notification
(#8418). Note that the migration to delete the now unused background
state has not been included, as it is non-essential and would have been
more difficult to backport. The migration to delete the unused state
will be included in the next major release instead.
This is a backport of #8314. Here's the original description:
MetaMask would sometimes get into a state where the notification popup
would never open. This could happen if the notification window was
closed shortly after being opened. After this happened, no popups would
show up until after the extension was reset.
This was happening because the background thought the popup was already
open. The variable it uses to track whether the popup was open or not
was being set to `true` immediately after the background asked the
browser to open a new window, before a handler was attached that could
respond to the window being closed.
Removing this line seems to solve the problem.
This line was added originally in #5437, which dealt with batch
transactions. Batches of transactions seem to work just fine without
this line though (from local testing), and I can't think of why this
would be required.
Closes#7051
* Version v7.7.7
* Update `@metamask/eth-ledger-bridge-keyring` (#8162)
* Update `@metamask/eth-ledger-bridge-keyring`
The Ledger keyring has been updated to ensure that any stale BIP44
accounts created prior to v7.7.6 of the extension are discarded when
the extension starts. Any attempts to sign with these accounts would
have failed; they needed to be re-added regardless.
* Update changelog
* Fix Ledger account index check for account zero (#8163)
Update Ledger keyring to fix bug when trying to sign with account 0
Co-authored-by: Mark Stacey <markjstacey@gmail.com>
* master: (34 commits)
Update changelog for v7.2.3
Fix e2e tests and gas default (#7267)
Do not transate on seed phrases
test:integration - fix renamed test data file
lint fix
test:e2e - fix bail condition
test:e2e - fix responsie argument
test:e2e - refactor missed spec file
test:e2e - only overwrite window.fetch once per session
test:e2e - rework fetch-mocks
test:e2e - add extra delay before closing popups
test:e2e - factor out prepareExtensionForTesting
test - e2e - dedupe fetchMocking + compose script as fn
Ensure correct tx category when sending to contracts without tx data (#7252)
Version v7.2.3
Add v7.2.2 to changelog
Update minimum Firefox verison to 56.0 (#7213)
Version v7.2.2
Update changelog for v7.2.1, v7.2.0, and v7.1.1
Add `appName` message to each locale
...
A new entry has been added for v7.2.1, and the Brave locales have been
moved from v7.1.1 to v7.2.0. That feature was mistakenly included under
the v7.1.1 heading - it was in fact released in v7.2.0
