Recommended way for extracting decimal components?

[jinchengJL]

Hello, I stumbled upon this library while looking for open-source C++ decimal implementations. It seems to be one of the few options with modern C++ support, and it looks very well maintained. Thanks for the great work.

One of the use cases I have is sending decimal values across the wire to other services. Ideally, the wire format can be a commonly understood binary representation of decimals, such as an integer mantissa and an exponent. While strings could also work, it would incur some extra serialization cost.

I wasn't able to find a way to extract the binary components of a decimal through the public interfaces of the library after some searching. So I'm wondering if there is a recommended way to do this? If not, would this be something you would consider supporting?

[mborland]

Hello, I stumbled upon this library while looking for open-source C++ decimal implementations. It seems to be one of the few options with modern C++ support, and it looks very well maintained. Thanks for the great work.

Thank you!

One of the use cases I have is sending decimal values across the wire to other services. Ideally, the wire format can be a commonly understood binary representation of decimals, such as an integer mantissa and an exponent. While strings could also work, it would incur some extra serialization cost.

I wasn't able to find a way to extract the binary components of a decimal through the public interfaces of the library after some searching. So I'm wondering if there is a recommended way to do this? If not, would this be something you would consider supporting?

So there's a couple ways you can do this.

1. frexp10 allows you to extract the significand and exponent directly: https://cppalliance.org/decimal/decimal.html#cmath_frexp10

2. If you are using decimal32 or decimal64 you could memcpy them into a std::uint32_t or std::uint64_t directly and then on the receiver side do a second memcpy to recover the value exactly

3. If you are using any of the types in the library you can use to_bid and from_bid (https://cppalliance.org/decimal/decimal.html#conversions) which extracts the bit pattern into a unsigned integer. This is nearly the same thing as 2 except more portable and applies to all 6 types provided.

Please let me know if that answers your question.

[jinchengJL]

Ah, excellent! frexp10 looks like exactly what I need. Don’t know how I missed it. Thank you, and closing this.