C++* < C++20

[porceCodes]

Big kudos. I deal with physical systems. I cannot overemphasize the importance of a proper use of units. I regularly come across the source codes that were written with ignorance of the unit compatibility, causing wreck-havoc only far later. I myself wrote proprietary solutions of this problem. It is immeasurably beneficial to have a standard way of dealing with this mater.

While I am a strong proponent of this work, I humbly ask you to consider that this work is too critical to be applied to C++20 only. The vast majority of code is written in C++11 and C++14. Do you see a path to have a support incumbent standards of C++?

Respectfully,

[mpusz]

This library is planned to become a part of the C++26 standard. This is why it has to benefit from the latest standards to provide the best possible interfaces for the future. I know that it limits the user base as of now but I hope more and more people will be able to use it in the upcoming months. As of now all of the major compilers already provide support for the library (https://mpusz.github.io/units/usage.html) so it should be possible to try it in nearly every environment.

[mpusz]

BTW, based on your experience from implementing and using a proprietary library, it would be great if you could share feedback so we can account for your requirements and have a better product in the future's standard.

[porceCodes]

will participate.

[grahamreeds]

I am in a similar position. Need a good library but working with GCC 9.3.0 and there is no plans to upgrade in the longer term. I would much prefer an actively maintained version rather than being stuck at v0.5.0 or writing our own.

How feature complete are previous versions (I am guessing things like speed, distance and size are all correct)? How reliant on cutting edge features are you? Would you be able to maintain support for older compilers longer by #ifdef and typedef'ing out things that are too cutting edge.

EDIT: It would also be nice if 0.5.0 was on Conan!

[chiphogg]

For users who need a good units library but can't support C++20, it's clear that mp-units will never be the answer (at least, not until they move to C++20). OK, so what is the answer for these users?

Here's a GitHub search query for dimensional-analysis-tagged libraries written in C++, ordered by stars. I'd say the best bet is to pick one with solid foundations and a strong community that meets your needs.

---

If you need guidance picking from this list, and you don't mind an obviously biased source 😁, I would recommend aurora-opensource/au, assuming you can support C++14. (My bias is that I'm the lead author and maintainer.)

Au is actively maintained, widely accessible, and almost trivially easy to install in any build system. We have extensive real-world production experience; fluent, composable APIs; an adaptive "overflow safety surface" for unit conversions with integral types; and fully unit-safe interfaces. We're the only units library I know of which delivers the kind of concise compiler errors which mp-units pioneered, outside of mp-units itself. And if you already have a units library --- or, you want to migrate away from Au later --- we're the easiest units library to migrate to or from.

---

Typically, I wouldn't presume to advertise my project on the repo of what is technically a competing project. 🙂 However, for this group of users, there's no competition; these are folks who simply can't use mp-units. What's more, my main answer gives an objective list of alternatives (so users can judge for themselves), and I've also stated my bias clearly and up front. I hope you find it helpful!

[chiphogg]

CC: @grahamreeds and @porceCodes, in case you were still looking for a solution.