We may have used it in our jenkins tests at some point but we don't
have those anymore. It weas pulling in ancient dependencies because
we were using version 2.0.0 which is fact much older than the
current version (1.2.0). We have little use for junit output anymore
so just remove it.
This allows you to increment an existing reaction below a message by clicking on
it.
At the moment, this is not linked to the action bar, so they each are using
local state. We'll likely want to add some mechanism so that we can local echo
to both of these UI areas at the same time, but that can be done separately.
Fixes https://github.com/vector-im/riot-web/issues/9486
This displays the existing reactions a message has from all users below the
message.
Since we don't currently have an API to actually get these events yet,
adds a temporary hook that looks for a specific message to inject some sample
data. This helps build out the UI for now and can be removed once it exists.
Fixes https://github.com/vector-im/riot-web/issues/9573
Concludes https://github.com/vector-im/riot-web/issues/8593
We are no longer seeing this error being triggered, and are considering it fixed. As a result, the dialog can be removed to reduce the amount of dead code in the project.
This adds the primary reactions to the action bar. They act as toggles where you
can only select one from each group at a time.
Note that currently we aren't actually sending the reaction at all. That's left
for a separate task.
Fixes https://github.com/vector-im/riot-web/issues/9576
This removes the clickable inviter behaviour, as it was too confusing to reveal
the user info sidebar and also hide the invite. Keeping both on screen would be
okay, but seems a bit too complex to resolve right before RC.
In addition, this adds the full inviter MXID to ensure it's clear who invited
you.
This applies the new design for multiple buttons in the message action bar,
paving the way for more things to appear here.
In addition, this changes the existing options button to use the three vertical
dots icon. Some theme colors are also tweaked to align with what they were meant
to be from the unified palette.
This adds a new action bar component to hold multiple per-message actions. This
existing options button has moved to this new component, and is currently the
only action.
This naming is clearer as it doesn't really edit at all (it shows a context
menu). This should also be less confusing with actual editing when it arrives.
Now that we have a fancier password complexity check, remove the older minimum
length to avoid the feeling of two password style guides fighting each other.
This adds a password complexity rule during registration to require strong
passwords. This is based on the `zxcvbn` module that we already use for key
backup passphrases.
In addition, this also tweaks validation more generally to allow rules to be
async functions.
In addition to migrating password fields, this also removes the remaining
support for old-style validation in registration now that all checks have been
converted.
When submitting a form, we want to validate more strictly to check for empty
values that might be required. A separate mode is used since we want to ignore
this issue when visiting a field one by one to enter data.
As an example, we convert the pre-existing logic for the username requirement
using this new support.
Since we're keeping the input as a ref anyway, let's use that rather than
requiring the value to be passed to `validate`. This allows others to call
`validate` as well.
This adjusts the submission step to focus the first invalid field and redisplay
validation. This also rearranges the older style field error handling on
registration which is slated for removal once we convert all fields to the new
style.
This updates the registration form to include the new-style validation state
when deciding whether the entire form is valid overall.
In addition, this tweaks the validation helper to take functions instead of
strings for translated text. This allows the validation helper to be create once
per component instead of once every render, which improves performance.
David Baker