Noteable - Accessibility Statement for the Noteable service (notebooks)

⤺ accessibility

Application accessibility statement in line with Public Sector Body (Websites and Mobile Applications) (No. 2) Accessibility Regulations 2018

This accessibility statement applies to the Noteable service described at (https://noteable.edina.ac.uk). Parallel statements are available for the website, moodle, and StatusPage.

This application is run by EDINA, for the University of Edinburgh. Our aim is for as many people as possible to be able to use this service. That means you should be able to:

• navigate most of the service using just a keyboard
• navigate most of the service using speech recognition software
• listen to most of the service using a screen reader (including the most recent versions of JAWS, NVDA and VoiceOver)
• understand the content of the application
• access and use the service without any time limits
• zoom in up to 200% without the text spilling off the screen
• zoom in and out while using the application
• change colours, contrast levels and fonts using your browser settings

Customising the application

AbilityNet has advice on making your device easier to use if you have a disability. This is an external site with suggestions to make your computer more accessible:

AbilityNet - My Computer My Way

With a few simple steps you can customise the appearance of our website using your browser settings to make it easier to read and navigate:

Additional information on how to customise our website appearance

How accessible this application is

We know some parts of this application are not fully accessible:

• some parts may not be fully compatible with screen readers (such as the Notebook Document pages)
• sometimes it can be hard to tell where you have navigated to using only the keyboard
• some non-text content does not have text alternatives
• some service features include pop-up windows. User are notified through the cursor changing to a hand-click icon when hovering over links which will open a new tab or window
• there may be some use of block italics and text over a background that changes colour
• Noteable is not primarily designed to be used on devices with particularly small screens smaller than 10 inches (such as a smartphone)
• some backgrounds change colour, with the user unable to pause or stop
• magnification beyond 200% may lead to reflow issues, where horizontal scrolling is necessary to see all content
• not all colour contrasts meet the recommended Web Content Accessibility Guidelines (WCAG) 2.1 AA standard
• some images may fail colour contrast guidelines if users redefine the colour palette in their browser

Feedback and contact information

If you need information on this website or the Noteable service in a different format such as an accessible PDF, large print, easy read, audio recording or braille:

• email edina@ed.ac.uk
• call us on 0131 650 3302
• British Sign Language (BSL) users can contact us via Contact SCOTLAND BSL, the on-line BSL interpreting service

We will consider your request and get back to you within 5 working days.

Reporting accessibility problems with this application

We are always looking to improve the accessibility of the Noteable service. If you find any problems not listed on this page, or think we're not meeting accessibility requirements:

• email us at edina@ed.ac.uk
• call us on 0131 650 3302
• British Sign Language (BSL) users can contact us via Contact SCOTLAND BSL, the online BSL interpreting service

We will consider your report and get back to you within 5 working days.

Enforcement procedure

The Equality and Human Rights Commission (EHRC) is responsible for enforcing the Public Sector Bodies (Websites and Mobile Applications) (No. 2) Accessibility Regulations 2018 (the 'accessibility regulations'). If you're not happy with how we respond to your complaint please contact the Equality Advisory and Support Service (EASS) directly.

Contact details for the Equality Advisory and Support Service (EASS)

The UK government has also produced information on how to report accessibility issues: Reporting an accessibility problem on a public sector website.

Contacting us by phone using British Sign Language

Contact SCOTLAND BSL runs a service for British Sign Language users and all of Scotland's public bodies using video relay. This enables sign language users to contact public bodies and vice versa. The service operates 24 hours a day, 7 days a week.

Technical information about this application's accessibility

EDINA is committed to making its Noteable service accessible, in accordance with the Public Sector Bodies (Websites and Mobile Applications) (No. 2) Accessibility Regulations 2018.

Compliance status

This application is partially compliant with the Web Content Accessibility Guidelines (WCAG) 2.1 AA standard, owing to the non-compliances listed below.

Non accessible content

The service is based on the Jupyter Notebook, and Notebooks are known to be complex web-applications.

Jupyter Notebook and Jupyter Lab applications

Project Jupyter has an over-arching accessibility working group which meets every 2 weeks.

The links to named issues are:

• Classic notebook Open Issues for Accessibility on GitHub
• Jupyterlab : Open Issues for Accessibility on GitHub

Notebook Documents have specific keyboard controls for the application, so the application does not fail 2.1.1 - Keyboard. There is a cheat-sheet for keyboard controls downloadable via data science cheat sheets.

Notebook Document code cells have specific keyboard controls within code-cells. There is a cheat-sheet for keyboard controls downloadable via DefKey.

RStudio

The RStudio interface interface uses an OpenSource application from RStudio. The company has no explicit Accessibility documentation for their product, however the following articles are reference material:

• A RStudio Accessibility Features (November 2022)
• A guide on setting up "RStudio Server" and a screen-reader (Sept 2021)

Non-compliance with the accessibility regulations

The following items to not comply with the WCAG 2.1 AA success criteria -

Classic Notebook

• 1.3.1 - Info and Relationships
  • There are no core landmark structures
  • The list of folders and files is styled like a table without a table element
  • The heading structure is not semantic or hierarchical
• 1.4.3 - Contrast Minimum
  • In code-cells, syntax highlighting may fail to have sufficient colour contrast
  • In Formgrader, the "breadcrumb" text does not have sufficient colour contrast to it's background
  • In Formgrader, the side-menu links do not have sufficient colour contrast
• 1.4.10 - Reflow
  • The notebook server page zooms to about 150% before adding a very short horizontal scroll bar
  • The notebook server page zooms to about 300% before adding a horizontal scroll bar and requiring scrolling
  • A notebook document page zooms to about 400% before adding a horizontal scroll bar and requiring scrolling
  • Both operate on a mobile phone without a horizontal scroll bar
• 2.1.2 - No Keyboard Trap
  • Code cells in notebook documents trap the tab key. As noted above, Both Notebook Documents and Code Cells use defined Keyboard Shortcuts for functionality, and the tab is not used to progress to the next focus element
• 2.2.1 - Timing Adjustable
  • All Notebook Server have timeouts:
    • an idle timeout of 60 minutes
    • a life-time timeout of 12 hours
  • Timeouts cannot be changed by the user
  • Users are not warned of impending timeouts
• 2.4.1 - Bypass Blocks
  • There is no "Skip to Content" link
  • There is no Landmark structure in the classic notebook interface
  • There is no Heading structure in the classic notebook interface
• 2.4.4 - Link Purpose (In Context)
  • The path breadcrumb trail in the notebook server page has an empty link when at the root level
• 2.4.6 - Headings and Labels
  • Checkbox for each item in filesystem list has no label
• 2.4.7 - Focus Visible
  • The focus indicator may not be particularly visible
• 3.2.2 - On Input
  • There is no indication when a link will be opened in a new window/tab. All notebooks, and most formgrader actions, are opened in a new window
• 3.2.3 - Consistent Navigation
  • On the notebook server page, the +GitRepo, Disk, and Empty Trash buttons do not always load in the same order
• 3.3.2 - Labels or Instructions
  • The checkboxes in the list of folders and files are both un-labelled and not contained in a form
• 4.1.2 - Name, Role, Value
  • The "select allquot; button in the files and folders list is a button with a role of checkbox, which contains another checkbox
  • The page changes from a file upload are not announced, and do not appear close to the upload button
  • In the Assignments tab, the elements that provide links the Noteable Documents / Feedback documents, the different assignments cannot be programatically identified

Jupyterlab Notebook

• 1.1.1 - Non-text Content
  • Not all inline SVG graphics have titles or labels
  • Images are used in Launcher "buttons". They generally do not have titles/labels
• 1.3.1 - Info and Relationships
  • There are no Landmarks to identify the structure and/or relationship of elements of the interface, however it does have a significant number of aria attributes
• 1.4.3 - Contrast Minimum
  • In code-cells, syntax highlighting may fail to have sufficient colour contrast
  • In Formgrader, the "breadcrumb" text does not have sufficient colour contrast to it's background
  • In Formgrader, the side-menu links do not have sufficient colour contrast
• 2.1.1 - Keyboard
  • Users cannot access all parts of the jupyterlab interface by keyboard alone:
    • The application menu is completely unavailable
    • Navigating between multiple documents in the working area is very difficult
    • Navigating between the side-bar and a document in the working area requires selecting the correct item in the side-bar/
  • There is a cheat-sheet for Notebook Documents at jupyter-notebook-shortcuts-pdf
• 2.1.2 - No Keyboard Trap
  • Code cells in notebook documents trap the tab key. As noted above, Both Notebook Documents and Code Cells use defined Keyboard Shortcuts for functionality, and the tab is not used to progress to the next focus element
• 2.2.1 - Timing Adjustable
  • All Notebook Server have timeouts:
    • an idle timeout of 60 minutes
    • a life-time timeout of 12 hours
  • Timeouts cannot be changed by the user
  • Users are not warned of impending timeouts
• 2.4.1 - Bypass Blocks
  • There is no "Skip to Content" link
  • There is no Landmark structure in the jupyterlab interface
  • There is no Heading structure in the jupyterlab interface
  • There are instances where components are loaded within an iframe
• 2.4.3 - Focus Order
  • The focus order does not include the application top-menu - just the side-bar and main work area
  • Focus disappears when tabbing up from the top of the side-bar
• 2.4.7 - Focus Visible
  • The focus indicator may not be particularly visible

RStudio Notebook

• 1.1.1 - Non-text Content
  • There are a few control images without alt-text or descriptive name
• 1.3.1 - Info and Relationships
  • There are incorrect aria-label attributes
  • There are Landmarks nested within other landmarks
  • Element IDs are not unique
• 1.4.3 - Contrast Minimum
  • In code-cells, syntax highlighting may fail to have sufficient colour contrast
  • Various parts of supplimental information in the interface may be insufficient colour contrast
• 2.1.2 - No Keyboard Trap
  • Whilst there is, technically, no keyboard trap - the interface uses a myriad of key-combinations to access various components, and the tab key can, does, get trapped in code-cells. The list of keyboard controls is available in the posit article Keyboard Shortcuts in the RStudio IDE
• 2.2.1 - Timing Adjustable
  • All Notebook Server have timeouts:
    • an idle timeout of 60 minutes
    • a life-time timeout of 12 hours
  • Timeouts cannot be changed by the user
  • Users are not warned of impending timeouts
• 2.4.1 - Bypass Blocks
  • There is no "Skip to Content" link - though main content is a difficult concept to define within the UI
• 2.4.7 - Focus Visible
  • The focus indicator may not be particularly visible
• 3.3.2 - Labels or Instructions
  • The checkboxes in the list of folders and files are both un-labelled and not contained in a form
  • Many inputs are javascript events on non-input events, so not programatically identifiable
• 4.1.2 - Name, Role, Value
  • A number of inputs are not identifiable, do not have name, and/or are not actually input elements

Disproportionate burden

We are not currently claiming that any accessibility problems would be a disproportionate burden to fix.

Content that's not within the scope of the accessibility regulations

We specifically exclude user-generated content, such as code run in Notebook Documents, from the scope of this report. For example, the addition of descriptions for graphs requires the user to include that in their code.

What we're doing to improve accessibility

The Noteable service team will work with the suppliers to address the accessibility issues. Unless specified otherwise, a complete solution or significant improvement will be in place for those items within our control by December 2023.

Preparation of this accessibility statement

The original statement was prepared on 20th October 2021. It was split into separate documents, and last reviewed on 2nd November 2022.

Testing

This service was last tested by the Noteable Service team within the Information Services Group directorate. The testing was carried out primarily using a combination of Firefox & Google Chrome browsers on a Linux system; with MacOS (and Safari) and Windows 10 (and Microsoft Edge) also used for comparative purposes, including screen reader compatibility. Lynx was used to test text-only navigation for the main website (but not the notebooks). In particular, the service was tested using Chrome on multiple platforms since it is by far the most popular browser when using a screen reader and the most popular browser when using the Noteable service. Current worldwide usage levels for different screen readers and browsers can be found in the May-June 2021 WebAIM annual survey.

Automated testing

We try to be consistent with automated testing across all components in the Noteable service.

A representative sample of pages were chosen to cover the different types of content and presentation within the service.

We used a range of toolkits and services to check each page:

• The IBM Equal Access Accessibility Checker is an open source tool for web developers and auditors that utilizes IBM's accessibility rule engine, which detects accessibility issues for web pages and web applications
• The Google Chrome Lighthouse DevTool is an open-source, automated tool for improving the quality of web pages. It has audits for performance, accessibility, best practice, progressive web apps, SEO and more
• The WAVE® Web Accessibility Evaluation Tool from WebAIM is a suite of evaluation tools that helps authors make their web content more accessible to individuals with disabilities
• The Accessibility tree viewer from the Firefox Developer Tools extension

We recognise the limitations of automated testing and we supplemented this with manual testing.

Manual Testing

Basic manual testing is viewing each page in a variety of browser/Operating System.

Content across the Noteable service has been tested without a mouse through keyboard navigation and has been found to be accessible using a keyboard interface. Note that the Jupyter and RStudio applications generally "tab" to navigate between sections, and "arrow key" within menus.

Using Windows 10 [Home Edition], We

• visually compared all pages using Firefox, Chrome, and Edge
• tested all pages using NVDA screen reader in Chrome and Edge (again, noting that the Jupyter and RStudio interfaces are complicated apps for a Screen Reader to address.)
• tested all pages using the Windows magnifier tool

Using MacOS X [Big Sur], we

• visually compared all pages using Firefox, Chrome, and Safari
• tested all pages isung the built-in voiceOver screen reader (again, noting that the Jupyter and RStudio interfaces are complicated apps for a Screen Reader to address.)

Using Linux [Ubuntu 20:04 LTS], we visually compared all pages using Firefox and Chrome.

We also tested for the following:

• spellcheck functionality
• scaling using different resolutions from –25% to +500%
• options to customise the interface (magnification, font, background colour et. cetera)
• notice of links to new window or tab when navigating the service
• information conveyed in colour or sound only
• Flashing or scrolling text
• time limits

Our findings on these features are summarised above.