OSCR

Accessibility

Research code should be readable by everyone who reads research. OSCR aims at the Web Content Accessibility Guidelines (WCAG) 2.2, level AA. It has not been audited against them, and does not claim to conform: this page says what it does and what is known not to work well yet.

The aim

Plain pages, built ahead of time, in one stylesheet, with the structure of a scientific archive: headings, lists, tables and words, rather than widgets and colours.

What the site does

  • Structure: the language of each page is declared; each page has one main heading, then headings in order; the masthead, the breadcrumb, the main content and the footer are landmarks; lists are lists and tables have headers.
  • Words before colours: a paper's status is said in words, never by a colour or a badge alone; the Code ↔ Paper reader lists every match in words, beside its colours.
  • Keyboard: every link, button and field can be reached and used with the keyboard, with a visible focus outline; the reader's panes can be focused and scrolled.
  • Forms: every field has a label; results and errors are announced to screen readers (role="status", role="alert"); nothing is sent before a review, for the removal request.
  • Zoom and small screens: the pages reflow down to a phone's width without scrolling sideways, and the browser's zoom enlarges everything; long names, addresses and titles wrap.
  • Contrast: text on the default ground meets a contrast of 4.5:1; the code's syntax colours keep 4.5:1 on every shade of the reader's matches.
  • Motion: the only animation, a loading bar, stops when your system asks for reduced motion.
  • Images: the logo and the badge have text alternatives; the site has no image that carries information without words.
  • Without JavaScript: every static page — a paper's record, the lists, the categories, the information pages — reads whole.

Known limits

  • The search, the DOI lookup, the pages of authors, journals, institutions, tools and datasets, the forms and the reader's paper pane need JavaScript.
  • The Code ↔ Paper reader — two panes that scroll on their own, colour-paired highlights — is the most complex part of the site and has not been tested with screen readers in depth; the list of matches and the file's source links are the accessible way through it.
  • The six colours of the matches are told apart by colour only on the page; their order and number are in the list of matches.
  • The paper's text in the reader is shown as plain text: its figures, tables and equations are not reproduced; the reader links to the paper.
  • The base text size is set in pixels: the browser's zoom enlarges it, but a default text size chosen in the browser's settings does not.
  • Content comes from papers and repositories as their authors wrote them: a title's formula may read awkwardly, a script may be uncommented.

Reporting a barrier

Open an issue on the project's GitHub, saying the page, what you tried to do, and what you use (browser, assistive technology). OSCR has no email address. The operator reads the project's issues; a fix goes into the public history of the code.