--------------------------------------
**Breaking changes in this release**:
* The *recursive* parameter of `.add()` is no longer ignored and now enforces its default value of `False` (explicitly set it to `True` for the previous behaviour)
* The glob pattern strings that may be passed to the `.add()` pattern parameter now actually behave like recursive glob patterns (see [the Python documentation](https://docs.python.org/3/library/glob.html) for how exactly)
* Most functions previously returning a dict with the raw JSON response, now return a custom mapping type instead
* This mapping type supports the original getitem syntax (`result["ItemName"]`) unchanged, but if you need an actual dictionary object you need to call `.as_json()` on it
* In the future response-specific subtypes with Pythonic accessors and object specific methods will hopefully be added
* HTTP basic authentication data to send to the API daemon must now be set as an `auth=(username, password)` tuple rather than using separate `username=` and `password=` parameters
Other changes:
* Added support for go-IPFS 0.5.x
* Adding directories with `.add()` has been greatly reworked:
* It's now possible to specify arbitrary rules on which objects to include a directory tree by passing a custom matcher object to the *pattern* parameter
* The new *period_special* parameter allows toggling whether glob patterns match dot-files implicitly and defaults to `True` (previously it was effectively `False`)
* The new *follow_symlinks* parameter similarly determines whether symbolic links will be followed when scanning directory trees and defaults to `False` (the previous default on Unix, albeit this likely wasn't intentional)
* `.add()` will now limit its scan to the directories required to match the given glob patterns (passing in regular expression objects will still scan the tree unconditionally however) – custom matchers have full control over which directories are visited
* The requests-based HTTP backend has been supplemented by another backend based on [HTTPx](https://www.python-httpx.org/) for Python 3.6+
* Due to a minor limitation within the library (no ability to apply address family restrictions during name resolution) this currently included as a preview and must be manually enabled, to do this ensure that the `httpx` library is installed in your Python environment and run your program with the environment variable *PY_IPFS_HTTP_CLIENT_PREFER_HTTPX* set to *yes*.
* In the hopefully not too long future, HTTPx will be used to finally provide async/await support for this library.