It is finally here! go-toml v2 is now stable. 🎉

Only just over one year after I started working on it. This version of go-toml has been rebuilt from the ground up. It respects the latest TOML specification. Countless bugs have been fixed. It behaves more like the standard library. And it is much faster. Special attention was given to keeping the public API to a minimum. Continuing the strict policy of backward compatibility, the next breaking change will be in another 5 years!

Thank you to everybody who contributed to this new version, as well as folks willing to try out the intermediate beta versions. Shout out to @moorereason for reporting many issues during the development process using differential fuzzing, as well as @bep for running a beta on the popular Hugo project!

There is still some work that could be done to improve the performance of the library (#758 and #669 for instance). But I am confident that the library is in a good-enough place at the moment for folks to reap the benefits of the rewrite. The major feature removal is the sort-of-AST that v1 provided. It allowed manipulating an arbitrary document and writing it back out. This was done to cut down the scope of the project, removing a feature that had unclear use-cases. The v2 parser creates some intermediate AST so that if somebody is up to design a Document API the tools should be available to do so. Please reach out if you are interested!

The readme contains a list of differences between v1 and v2 to be aware of when upgrading. Please report any issue you may have with the new version in the bug tracker 🐞, provide feedback, and ask questions in Discussions 💬.
With this release, go-toml v1 will not be receiving any updates. This is required to keep maintenance of the project to a minimum.

Hopefully, this change doesn't break your thing!

Full commit log: c9a09d8...v2

Thank you contributors: @pelletier @oschwald @moorereason @xwjdsh @kkHAIKE @RiyaJohn @jidicula @zostay @mmorel-35.