As brought up on #782, there is an asymetry between numbers go-toml can encode
and decode. Specifically, unsigned numbers larger than math.Int64 could be
encoded but not decoded.
We considered two options: allow decoding of those numbers, or prevent those
numbers to be decoded.
Ultimately we decided to narrow the range of numbers to be encoded. The TOML
specification only requires parsers to support at least the 64 bits integer
range. Allowing larger numbers would create non-standard TOML documents, which
may not be readable (at best) by other implementations. It is a safer default to
generate documents valid by default. People who wish to deal with larger numbers
can provide a custom type implementing encoding.TextMarshaler.
As brought up on #782, there is an asymetry between numbers go-toml can encode and decode. Specifically, unsigned numbers larger than
math.Int64could be encoded but not decoded.We considered two options: allow decoding of those numbers, or prevent those numbers to be decoded.
Ultimately we decided to narrow the range of numbers to be encoded. The TOML specification only requires parsers to support at least the 64 bits integer range. Allowing larger numbers would create non-standard TOML documents, which may not be readable (at best) by other implementations. It is a safer default to generate documents valid by default. People who wish to deal with larger numbers can provide a custom type implementing
encoding.TextMarshaler.Refs #781
cc @jmank88