This relies on the behaviour of pandoc, and as such may not apply to
other Markdown renderers.
Before this change, you would have something like:
> - Example: `application/json`
>
> ```javascript
> "HELLO, HASKELLER"
> ```
When converting this to HTML, PDF, etc. the code block is _not_
contained within the bullet point.
With this change, the generated markdown looks like:
> - Example: `application/json`
>
> ```javascript
> "HELLO, HASKELLER"
> ```
With pandoc at least, this effectively indents the entire code block
to be under the bullet point, which is the intended effect.
Note that the code itself is _not_ indented (which might break other
Markdown renderers) as to do so would require splitting on newlines,
which may have unintended consequences when dealing with generated
values (may contain `\r\n`, etc.).
Currently, it appears after the notes, authentication and capture
information... such that if any of those exist, then the list of
header sensitivity will appear to be an item of one of the previous
ones (as they provide Markdown headers).
As reported in #754, `HasDocs` instance of `ReqBody` was dropping samples other
than the first one. With this patch we show at most `_maxSamples` samples for
`ReqBody`, and also include the sample title in the docs.
We use NoContent to signify an empty response nowadays. This commit
replaces all occurences of () with NoContent so that all packages use
the new semantics.
This introduces a `Delayed` type in `RoutingApplication.hs` that
contains a handler together with delayed checks. There are several
blocks of delayed checks, so that we can ultimately execute them in the
order we desire.
The process is documented in more detail in `RoutingApplication.hs`.
With DocOptions one can limit maximum number of samples.
This is useful for Generic-based instances for recursive data types
(e.g. `[]`). Default options set maxSamples to 5.
