12.4. Design Doc 003: ToS rendering

12.4.1. Summary

This document describes how terms of service (ToS) as well as other “legal agreement documents” are served, represented and rendered.

12.4.2. Motivation

Different exchanges and backup/sync providers each have their custom legal agreement documents. As we don’t know all providers and they are not centrally registered anywhere, these documents can’t be hardcoded into wallet applications. Instead, these service providers expose endpoints that allow downloading the latest version of these legal agreement documents.

These documents must be rendered on a variety of platforms in a user-friendly way.

12.4.3. Proposed Solution

The service providers can output legal agreements in various formats, determined via the "Accept: " request header. The format provider must support the text/plain mime type. The format provider must support the text/markdown mime type. Except for styling and navigation, the content of each format of the same legal agreement document should be the same.

Legal documents with mime type text/markdown should confirm to the commonmark specification.

When wallets render text/markdown legal documents, they must disable embedded HTML rendering. Wallets may style the markdown rendering to improve usability. For example, they can make sections collabsible or add a navigation side-bar on bigger screens.

It is recommended that the text/markdown document is used as the “master document” for generating the corresponding legal agreement document in other formats. However, service providers can also provide custom versions with more appropriate styling, like a logo in the header of a printable PDF document.

Markdown Conventions

The text/markdown document should follow the commonmark spec. Main headlines (level 1) and their following content (until the next main headline) will be shown as expandable sections in wallets.

The document must begin with a main headline:

# First Headline

or

First Headline
==============

12.4.4. Alternatives

We considered and rejected the following alternatives:

  • Use only plain text. This is not user-friendly, as inline formatting (bold, italic), styled section headers, paragraphs wrapped to the screen size, formatted lists and tables are not supported.
  • Use HTML. This has a variety of issues:
    • Service providers might provide HTML that does not render nicely on the device that our wallet application is running on.
    • Rendering HTML inside the application poses security risks.
  • Use a strict subset of HTML. This would mean we would have to define some standardized subset that all wallet implementations support, which is too much work. Existing HTML renderers (such as Android’s Html.fromHTML) support undocumented subsets that lack features we want, such as ordered lists. Defining our own HTML subset would also make authoring harder, as it forces authors of legal agreement documents to author in our HTML subset, as conversion tools from other format will not generate output in our HTML subset.
  • Use reStructuredText (directly or via Sphinx). This at first looks like an obvious choice for a master format, since Taler is already using reStructuredText for all its documentation. But it doesn’t work out well, since the only maintained implementation of a parser/renderer is written in Python. Even with the Python implementation (docutils / Sphinx), we can’t convert .rst to Markdown nicely.

12.4.5. Drawbacks

  • Markdown parsing / rendering libraries can be relatively large.

12.4.6. Discussion / Q&A

  • Should the legal agreement endpoints have some mechanism to determine what content types they support?