The e2e test script now accepts a `--retry-until-failure` flag that can
be useful for locally reproducing intermittent failures.
Normally the `retries` option will only retry upon failure. But if
you want the test to fail to inspect the failure state, you want it to
keep running until the first failure then stop. This flag accomplishes
that, reversing the retry conditions.
Our e2e test driver was building the locator object multiple times over
in some cases. The `Locator` object is required by certain webdriver
methods, but our driver methods all accept a "raw locator".
This change avoids a few redundant calls, and makes it easier to
improve the error message for a timeout failure (which will be done in
a later PR). It also makes the code match the documentation/parameter
names.
The default network controller state is currently invalid on test
builds. The initial state is set to localhost (Ganache) on test builds,
but that network configuration is missing.
The initial state for test builds has been updated to include the
network configuration for localhost. Additionally, the initial state
was updated to only be applied if persisted state is missing. This is
to ensure the initial network configuration state doesn't override test
fixtures in e2e tests.
The legacy gas API is still useful for BSC, which is a network our APIs
support that is not EIP-1559 compatible. The legacy gas API will now be
used for BSC prior to using RPC methods as a fallback.
This brings extension closer in alignment with mobile, which also uses
the legacy gas API for BSC.
The E2E network mock function has been updated to use a variable for
the initial test network. This made it easier to write the e2e test for
the BSC case.
Two new security advisories have been resolved. These advisories are
causing CI to fail on `develop`. Neither presents any risk to us,
as they are prototype pollution issues that are prevented by lockdown.
The first advisory isn't easy for us to patch. It's caused by an
outdated version of `protobufjs` used by `@trezor/transport`. It has
been ignored for now, until Trezor updates that package.
For the second advisory (related to `tough-cookie`), it was resolved
by updating that dependency in our lockfile.
In the new version of NetworkController, it will now precreate network
clients for built-in and custom networks and expose those network
clients for consumers. This furthers the multichain UX project by making
it possible for MetaMask to interface with multiple networks
simultaneously.
This commit also upgrades `@metamask/gas-fee-controller` to prevent a
peer dependency warning from showing up as well as
`@metamask/controller-utils` in order to reduce the dependency tree.
There are no user-facing changes to either package.
* feat: adds linea mainnet support in anticipation of July 11 launch
* chore: fixes linting issue
* chore: adds comma for linting
---------
Co-authored-by: Pedro Pablo Aste Kompen <wachunei@gmail.com>
Co-authored-by: Dan J Miller <danjm.com@gmail.com>
* Add support for snap authorship component at the top of PermissionConnect
* Add PermissionCellOptions
* Add details popover
* Add action for revoking dynamic permissions
* Improve UI and revoke logic
* Better eth_accounts screen support
* Fix tests
* Fix lint
* More linting fixes
* Fix missing fence
* Add another fence
* Unnest permission page to fix weird CSS issues
* Hide footer on permissions connect when using a snap
* Stop estimating gas when starting a draft transaction in the send flow, because other estimates will happen later
* fix: test fixing overlay dom issue
* Stop estimating gas when starting a draft transaction in the send flow, because other estimates will happen later
* fix: remove extra await delays
---------
Co-authored-by: Danica Shen <zhaodanica@gmail.com>
The `metamaskbot` account now comments with a failure message when the
policy update fails. This comment also includes a link to the specific
run that failed, so that the PR author can review the log and/or retry
the policy update.
* update button font weight to medium
* update design-tokens and add new Text medium
* undo button base weight change
* fix unused import
* update snapshot
* fix yarn dedupe
* update readme
Today if you try to re-run failed jobs in a policy update, it will
fail to restore caches from jobs that were successful. This could
result in a partial update, or a misleading "No policy changes"
message when there are policy changes to make. It's only safe to re-
run the entire workflow.
The cache key has been updated to reference the commit hash instead of
the run ID, making it safe to re-run just the failed jobs.
