We’ve seen in letters usability testing that people get stuck in a
“no-man’s land” when trying to go back from the _Send yourself a test_
page.
This was broken for two reasons:
- we hadn’t considered that a letter template without placeholder still
requires you to fill in
- we’ve changed subsequently made the _view template_ page the place
where you do your actions, rather than the (old) page with all the
templates shown
So this commit fixes it so that the back link always take you back to
the page you were previously on, and adds some more test cases so we
have all the scenarios accounted for.
Address line 3 is optional. Currently the only way we have of indicating
to users what is/isn’t optional is by using the example. Which probably
isn’t ideal, but should at least be correct.
The breaking change page wasn’t properly accounting for the fact that
letter recipients span multiple columns – it was assuming they’d only
take up one column like they do for email and SMS.
This commit fixes:
- the number of column headers (A, B, C, …) to be correct
- the count of columns (you will need X columns in your file) to be
correct
It then parameterises the test to look at a case where a recipient is
in one column (email) and multiple columns (letter).
Users who go to edit the contact details for a letter from the template
page get very confused when they click save and are dumped on the
settings page. It doesn’t match the way editing other parts of
letter works, and you can’t see an accurate preview of the changes from
the settings page.
So this commit changes the flow to go from the _edit contact details_
page back to the _view template_ page when the user has got there by
clicking the blue _Edit_ button on the _view template_ page.
pass in the base URL - if not set in the environment this is set to
localhost, but on paas we can pull this out of vcap_services so that
letters render properly on paas
Tests assumed that the API returns the template `id` as part of the
object. It doesn’t – it returns it as the key used to look up the
object. The `id` was missing from the transformation into the format
used by the front end.
For some reason Flask is fine building the URL with `template_id=None`,
but obviously this doesn’t generate a valid link.
When adding a letter it’s hard to know how the stuff you’re typing is
going to look. So rather than go straight to the edit form let’s show
users a blank letter. They can then choose which part of it they want
to edit first. And will have a better idea of how their changes are
going to show up.
In research we also saw nervousness around saving a template and wanting
to ‘preview’ it first. Hopefully this flow will make people feel less
precious about saving a template because they’ve already done it once
just by creating the template.
When you add a new template the next thing you want to do is probably:
- send a test
- edit it after seeing what it looks like
So let’s redirect you straight to the page where you can do these
things.
This is especially important for letters where we want to tighten the
edit/feedback loop.
Not necessary to have it on its own page – it’s one line of stuff. And
definitely not as frequent use as the ‘Upload recipients’ or ‘Send
yourself a test’ links.
Because we have space there’s no need to hide this now. Makes it less
confusing to have parts of the email appear/disappear as you’re clicking
through the flow.
Not everyone knows how to use `ctrl` + `f`, and it’s not scoped to
just the list of templates.
The template you want to work with is often not the first one in the
list, but ordering by created at is useful for other reasons (mainly
around first time use).
This commit adds a find as you type control which aims to give users a
quick way of getting to the template they want to work with.
When a team has lots of templates the choose template page gets very
long. It gets hard to find the template that you are looking for.
Our initial reckon was that teams would not be giving their templates
very useful names, and therefore a preview would be helpful. What we
have found is that:
- teams actually do give their templates useful names, and refer to
these template names elsewhere
- the previews are less useful for emails and text messages, because
they have so much content (which for emails also makes it harder to
`ctrl` + `f` the template name)
The other problem we found was that this page presented the user with
a _lot_ of options. For each template there were 4 actions, plus the
click-to-preview action for letters, plus the ‘see previous version’
action for templates that had been edited multiple times. It was a very
busy page.
And the final problem (that we recently introduced) was that there was
no way, other than the visual cues, to know whether a template was a
letter, email, or text message.
So this commit strips back the choose template page to be very focused
on finding the right template, by only showing the template name and
type. The user can then click through to a page that shows just a single
template, and perform actions relevant to that template from that page.
The tour teaches you how Notify works by letting you do a thing and then
showing you the effect of the thing you’ve just done – a text message on
your phone.
This is not as effective if you don’t get the text message quickly. It
breaks the association you make between what you’ve done on the computer
and what’s happening on your phone.
Slow text message delivery can happen if you’re doing a big job. We can
get around this by making your text message use the priority queue.
This was observed in the pilot research session yesterday.
The breaking change page temporarily holds the changes in hidden inputs
on the page. The messages content it gets from the `.content` property
on the subject. This is raw and not transformed in any way, so fine.
For the subject it gets the value from the `.subject` attribute on the
template. For email templates, this will be transformed to highlight
placeholders with `<span class='placeholder'>…`. This means that when
the change is confirmed, it’s this encoded version that gets sent to the
API. Which is bad, because we then save `<span class='placeholder'>` in
the database.
This commit changes the page to look at the `._subject` attribute
instead, which is the internal, untransformed version of the subject.
The support flow was using `yes` and `no` to mean emergency/not
emergency. But not in all places – in one place it was using
`True`/`False` instead.
We were treating anything other than `yes`/`no` as a non-answer, which
means ask the question again. Because of the `True`/`False` thing, there
was no way of the user providing a valid `yes`/`no` answer. Which means
that we just kept asking them the question again and again and they got
stuck in a loop.
For some reason we were rebuilding `new_template` as a dictionary,
without the `placeholders` attribute. This meant that we were never
actually counting the placeholders, just counting the length of `None`
and adding 1 to it.
So this commit fixes that, beefs up the tests, and makes sure that
everything is pluralised properly.
no actual template functionality yet - just the ability for services
that have letters enabled to edit a 10 line block that will go on the
top right hand side of their letters with contact information
Right now we have separate pages for email and text message templates.
In the future we will also have a separate page for letter templates.
This commit changes Notify to only have one page for all templates.
What is the problem?
---
The left-hand navigation is getting quite crowded, at 8 items for a
service that can send letters. Research suggests that the number of
objects an average human can hold in working memory is 7 ± 2 [1]. So
we’re at the limit of how many items the navigation should have.
In the future we will need to search/sort/filter templates by attributes
other than type, for example:
- show me the ‘confirmation’ templates
- show me the most recently used templates
- show me all templates containing the placeholder `((ref_no))`
These are hypothetical for now, but these needs (or others) may become
real in the future. At this point pre-filtering the list of templates
by type would restrict what searches a user could do. So by making this
change now we’re in a better position to iterate the design in the
future.
What’s the change?
---
This commit replaces the ‘Email templates’, ‘Text message templates’ and
‘Letter templates’ pages with one page called ‘Templates’.
This new templates page shows all the templates for the service, sorted
by most recently created first (as before).
To add a new template there is a new page with a form asking you what
kind of template you want to create. This is necessary because in the
past we knew what kind of template you wanted to create based on the
kind you were looking at.
What’s the impact of this change on new users?
---
This change alters the onboarding process slightly. We still want to
take people through the empty templates page from the call-to-action on
the dashboard because it helps them understand that to send a message
using Notify you need a template. But because we don’t have separate
pages for emails/text messages we will have to send users through the
extra step of choosing what kind of template to create. This is a bit
clunkier on first use but:
- it still gets the point across
- it takes them through the actual flow they will be using to create new
templates in the future (ie they’re learning how to use Notify, not
just being taken through a special onboarding route)
I’m not too worried about this change in terms of the experience for new
users. Furthermore, by making it now we get to validate whether it’s
causing any problems in the lab research booked for next week.
What’s the impact of this change on current services?
---
Looking at the top 15 services by number of templates[2], most are using
either text messages or emails. So this change would not have a
significant impact on these services because the page will not get any
longer. In other words we wouldn’t be making it worse for them.
Those services who do use both are not using as many templates. The
worst-case scenario is SSCS, who have 16 templates, evenly split between
email and text messages. So they would go from having 8 templates per
page to 16, which is still less than half the number that HMPO or
Digital Marketplace are managing.
References
---
1. https://en.wikipedia.org/wiki/The_Magical_Number_Seven,_Plus_or_Minus_Two
2. Template usage by service
Service name | Template count | Template types
---------------------------------------|----------------|---------------
Her Majesty's Passport Office | 40 | sms
Digital Marketplace | 40 | email
GovWifi-Staging | 19 | sms
GovWifi | 18 | sms
Digital Apprenticeship Service | 16 | email
SSCS | 16 | both
Crown Commercial Service MI Collection | 15 | email
Help with Prison Visits | 12 | both
Digital Future | 12 | email
Export Licensing Service | 11 | email
Civil Money Claims | 9 | both
DVLA Drivers Medical Service | 9 | sms
GOV.UK Notify | 8 | both
Manage your benefit overpayments | 8 | both
Tax Renewals | 8 | both
The user has 10 tries at the password, after which the account is locked.
The same is true for the verify code, the user will have 10 tries before the user account is locked.
> When the CSV is missing the header row, we get an error and the user
> will see "Sorry, we are experiencing technical difficulties..."
>
> We should return a better error message for the user.
– https://www.pivotaltracker.com/story/show/140668615
This was caused by an attempt to access the `first_recipient` variable
before it was assigned. It would only be assigned when there was at
least one row in the file.
Fixing this means doing two things:
- defaulting `first_recipient` to be `None` before looking in the file
- adding an error message for when we can’t extract any rows out of the
file (which is more nuanced than the file just being completely empty)
(There’s a nasty `sort` in the Jinja template because when there are no
rows in the file the order of the required column headers is not
deterministic.)
specifically, the 2FA page when you first create an account is different to the login 2FA page
and also the 2FA page when you change your phone number is different as well
when a user enters their 2FA code, the API will store a random UUID
against them in the database - this code is then stored on the cookie
on the front end.
At the beginning of each authenticated request, we do the following
steps:
* Retrieve the user's cookie, and get the user_id from it
* Request that user's details from the database
* populate current_user with the DB model
* run the login_required decorator, which calls
current_user.is_authenticated
is_authenticated now also checks that the database model matches the
cookie for session_id. The potential states and meanings are as follows:
database | cookie | meaning
----------+--------+---------
None | None | New user, or system just been deployed.
| | Redirect to start page.
----------+--------+---------
'abc' | None | New browser (or cleared cookies). Redirect to
| | start page.
----------+--------+---------
None | 'abc' | Invalid state (cookie is set from user obj, so
| | would only happen if DB is cleared)
----------+--------+---------
'abc' | 'abc' | Same browser. Business as usual
----------+--------+---------
'abc' | 'def' | Different browser in cookie - db has been changed
| | since then. Redirect to start
> Users that allow their session to expire, or access a bookmarked link
> are told they need to "Sign in to access this page" - we should
> explain that it's because they've been away a while, so that they
> understand why they're being asked to log in again.
– https://www.pivotaltracker.com/story/show/140016919
The message we were showing before (Please log in to access this page is
the default message from Flask Login).
In order to stop this flash message from appearing, we need to override
the default handler for when a user is unauthorised. We’re overriding it
with the same behaviour, minus the flash message.
If you navigate deliberately to the sign in page it’s unchanged.
Content is Sheryll-approved.
Matches what we do on the usage page. No need to see months in the
future because there’s no way you’ll have sent any messages in those
months, unless you’re Marty McFly.
The previous, weekly activity breakdown was what we reckoned might be
useful. But now that we have people using the platform it feels like
aggregating a service’s usage by month is:
- matches the timeframe users report on within their organisation
- is consistent with the usage page
And like the usage page this commit also limits the page to only show
one financial year’s worth of data at once (rather than data for all
time).
This commit also makes some changes to the jobs view code so that our
aggregation of failure states is consistent between the dashboard pages
and the jobs pages.
