With the Guest registration calling policyactions.js the [data-action="view"]
was having two click handlers set on the inital policy modal trgger.
With this commit we state what we want the click event to be set on.
If javascript is disabled, we don't need to open the menus at all. If it is enabled
we should use the custom step where possible.
In some tests (user profile) we have to use link_or_button because
the element that opens the menu will be a link or a button in different themes.
When accepting the policies via the sitepolicy handler, only compulsory
policies are to be marked as accepted. Optional policies will be left as
pending. Users must express their consent explicitly for them.
As a side product of the change, unit tests are added for the whole
handler class.
This adds support for optional policies to the user acceptance reports.
Distinguished are "Pending" acceptances (we did not hear yet) from
"Declined" (user did not agree). The status workflow updated to support
new transitions: pending -> declined and declined -> accepted.
There was an accessibility issue with the previous icons that we used
the same "checked" shape just in different colours for different
meanings. New icons added for the new statuses:
* partial - a warning icon for the overall status column that the user
has only some policies accepted, not all.
* pending - that we did not hear yet from the user - whcih is different
from a declined policy.
For optional policies, we provide a radio selector to let the user
choose the acceptance status on the consent page. For policies displayed
on their own page, we display a link to decline the policy.
The way how we pass the list of policy version ids to index.php has
changed so that we can now not only pass the list of ids, but also the
actual acceptance status (accepted / declined).
This method allows to quickly check if the given policy version is
marked as optional or compulsory. This will be needed in other places
such as permissions check.
The method now returns three-state logic. A bool value true/false is
returned if the user has accepted/rejected the policy, respectively. A
null value is returned if the user did not express their agreement in
either way yet.
This allows to distinguish between "rejected the policy" and "did not
say anything about it yet" cases.
The patch adds a new column to the database table to hold the
information if giving agreement to the policy is compulsory or optional.
The flag can be defined via the policy editing form and is displayed at
the policies management screen.
The last modified time merged with the version column at the policies
management screen to save a bit of horizontal space - needed as we
display more information now at the first column.
The tests make sure that both existing users as well as signing up users
can accept policies with "on their own page" style of agreement prior
reaching the consent page.
The idea of the patch is to check the list of policies that are shown
before and on the consent page. If that list contains a policy that is
supposed to be accepted on its own page, the existing flow is
interrupted and the user is redirected to a view.php to display that
particular policy.
The view template for such a policy contains a button for getting the
user agreement. When pressed, the policy is marked as accepted and we
can go to start again.
To make this working during the signup, we need to extend the existing
logic and start tracking which particular policies the visitor accepted
prior reaching the signup form. Similarly, we need to start track which
particular policies were displayed and use that list when evaluating
which policies were unchecked on the consent page.
The patch introduces a new property of a policy document called
"Agreement style". It allows the admin to define if the policy should be
accepted together with all others on the consent page (legacy and
default behaviour) or on its page before the consent page is reached
(the new optional behaviour).
Floating banners cause issues with clickability in Behat as it is unable
to understand that it cannot interact with the elements underneath the
floating banner, or that it needs to scroll the page so that the
required content is no longer beneath the floating banner.
Changing the banner to be fixed to the bottom of the page during Behat
runes is a reliable fix.
The way in which the modal was displayed meant that there were no
pending JS items, whilst the modal was rendered, causing random behat
fails.
This JS has been restructured to create the Modal and pass it a set of
Promises for both the title, and body.
