- proto::parser will likely contain helper functions for parsing
messages from a buffer, for now it will explicitly define message
parsers for each available message. It also leverages the new unpacker
API, which has type safety in mind. These messages should not be
unstructured, so it doesn't make sense to use the token API.
msgpack::map_view can be used to iterate, pair-wise, over a range of
msgpack::token. It will immediately return if the first token is not a
map, and will skip over nested map/arrays.
Note for the future: It will be handy to be able to get the subspan
corresponding to the nested map/array. Will think about how to solve
that later.
Begin incorporating map_view into the server.
Add formatters for std::byte, dynamic theme for bool, and spans thereof.
Maybe switch to range?
- Added token_reader::read_some(), which takes a view of a token buffer
reads/unpacks up to that many tokens from a buffer. If an error occurs,
or if there are fewer msgpack formats within the buffer, then the
returned span will be a subspan of the original buffer. If no tokens
were read before encountering an error, token_reader will return the
error instead.
- token::operator== overload for types that might be stored within the
token. Handy for cleaning up tests, even though both methods of
`token::get<T>() == value` and `token == value` are represented for
now.
- Added a simple tool to open a file containing msgpack content and read
tokens into a buffer. This is for testing speed, but needs to be made
into a proper benchmark once more work has been done. Pretty fast
right now, only taking 2us to parse ~200 tokens (that takes the
pythong msgpack implementation ~1s to handle).
This essentially completes the token_reader. In the future, we will
probably want something that lets us allocate maps/arrays more
seamlessly, but we can start working on client-server architecture with
this reader.
