This should have no effect since all (or most) existing users would be migrated with the last 2 major releases and now only new users would be affected
* Migrate crypto widgets visibility in NTP
fix https://github.com/brave/brave-browser/issues/19708
Migration is done once in the first NTP loading.
For new users, all crypto widgets are hidden by default.
For existing users, if current foremost widget is crypto widget,
only it is visible and others are hidden.
And, authed widgets are visible regardless of its position in widget stack
After migration, user can control crypto widget's visibility from gallery.
To make all crypto widgets hidden by default, each pref's default
value is not changed. Their default prefs is still "true".
As their visibility are managed by prefs and widget order is managed in local
storage, it's difficult to start migration from c++. If default is changed
to false, it's very difficult to determine whether foremost widget is crypto
or not in the NTP webui.
Previously, the front-end would decide on the order and type of card used for each feed item. Having the mojom API return that makes it easier to have the same decisions apply to desktop and android.
Also uses strong typings from mojom es module output. Works very well with Typescript.
Now that the mojom API is feature-complete, we can remove the extension API.
fix https://github.com/brave/brave-browser/issues/18488
Launched dialog for 3rd, 10th and 20th visits for brave search page
when user doesn't enable web discovery option yet or doesn't check
don't ask checkbox before.
fix https://github.com/brave/brave-browser/issues/18305
Selected VPN region data is stored as dict in prefs.
Moved existing kBraveVPNShowButton pref name to brave_vpn/pref_names to
manage vpn related prefs in one place.
This reverts commit 0b78f9e1fc, reversing
changes made to d260b6aa38.
Reverting because upstream has fixed the first run regression via:
https://chromium.googlesource.com/chromium/src.git/+/f9385e0d893916cbb16b477bdec127e56f9897da
Cherry-picked into cr94 as
https://chromium.googlesource.com/chromium/src/+/3f20e2edeb0adf78a6a9a6208e2e02ad3c7d8810
commit f9385e0d893916cbb16b477bdec127e56f9897da
Author: Caitlin Fischer <caitlinfischer@google.com>
Date: Mon Aug 30 13:31:02 2021 +0000
Ignore Local State when deciding whether to show the first run dialog.
The bug was caused by an experiment that causes Local State to be
written to disk during startup (before the first-run-dialog-showing
logic is reached) on some clients. Since Local State existed on some
clients, the dialog was not shown. However, the first-run-dialog-showing
logic should not depend on the presence or absence of a Local State
file.
This change affects some POSIX platforms running Google Chrome for the
first time. This logic being removed was originally added here:
https://chromiumcodereview.appspot.com/14946003/.
Bug: 1221483
Chromium change:
https://bugs.chromium.org/p/chromium/issues/detail?id=1223264https://chromium.googlesource.com/chromium/src/+/7c683fe7de2aaff98f1ac8f0c2a24d2a44635499
commit 7c683fe7de2aaff98f1ac8f0c2a24d2a44635499
Author: Peter Kasting <pkasting@chromium.org>
Date: Thu Jun 24 15:12:10 2021 +0000
Fix -Wloop-analysis warnings in chrome/.
These indicate when a range-based for loop is using an index whose type
(value, pointer, or reference) doesn't match what the loop actually
extracts from the range. Fix by matching the actual type better.
This shouldn't cause any behavior/performance change, just be slightly
clearer about what's actually happening when reading the code.
Bug: 1223264
Mario Sanchez Prada