commit | 33034139f07dbc26ff4ef861bd9100b5b03f67bb | [log] [tgz] |
---|---|---|
author | Olaf Bergmann <bergmann@tzi.org> | Fri Apr 24 17:12:40 2015 +0200 |
committer | Olaf Bergmann <bergmann@tzi.org> | Fri Apr 24 17:12:40 2015 +0200 |
tree | ee8f4d8ea75cbbf7a5aaa6348e69c98fbdec6700 | |
parent | b444590950ec8959e696640c86450331f5f17e3d [diff] |
Optional float encoding/decoding When CBOR_NO_FLOAT is defined at build time, support for floating point types is switched off. When trying to parse a data item of type 7 and additional information 25 to 27 (i.e., items starting with 0xf9 to 0xfb), the parser will return the new error code CN_CBOR_ERR_FLOAT_NOT_SUPPORTED.
This is a constrained node implementation of CBOR in C that I threw together in 2013, before the publication of RFC 7049, to validate certain implementability considerations.
Its API model was inspired by nxjson. It turns out that this API model actually works even better with the advantages of the CBOR format.
This code has been used in a number of research implementations on constrained nodes, with resulting code sizes appreciably under 1 KiB on ARM platforms.
I always meant to improve the interface some more with certain API changes, in order to get even closer to 0.5 KiB, but I ran out of time. So here it is. If I do get around to making these changes, the API will indeed change a bit, so please be forewarned.
There is a Simple-Makefile
for playing around, as well as a complete cmake
-based build environment. (You can choose what fits your needs better.)
Building with cmake
:
./build.sh
Building including testing:
./build.sh all test
Generating a test coverage report (requires lcov[^1]; result in build/lcov/index.html
):
./build.sh all coveralls coverage_report
License: MIT
[^1]: Installation with homebrew: brew install lcov