Open legendre6891 opened 1 month ago
First of all, thanks a lot!
Actually, I do not plan to polish this project into something really usable by other people. It's more of a "playground" thing. (Rather, I plan to insert at some point in the future such a polished version into my long-abandoned project fp
when I finally work on reincarnating it) The primary goal of this repo is to experiment things and do benchmarks. I chose the e
-format because that's what Ryu-printf (the main target of comparison) did, and Ryu-printf did so probably because its main focus is on generating large number of digits of general floating-point numbers, where the e
-format is much more natural than the f
-format.
So, no, I do not consider it "necessarily in-scope" for this project alone. But in a broader context, of course I do consider it, and obviously it's needed for any practical applications. As I said, I will (hopefully) at some point write an implementation and merge it into fp
. In fact, I already have written one when I contributed into Boost.CharConv, based on its original author's work: https://github.com/boostorg/charconv/pull/169. But generally I'm not very happy with std::to_chars
interface spec, so I'll probably write a different one later.
In any case, if you're considering writing one by yourself, I cannot encourage you more. If you're planning to make a PR with that, that would be even greater.
Or, if you're just looking for a usable implementation, I recommend you to go with Boost.CharConv or {fmt}
. Boost.CharConv's implementation is based on this project, and {fmt}
's implementation essentially contains some portion of this project.
Hi -- this is a very interesting technical project!
I wonder if
%.[prec]f
-style formatting is considered in scope? It seeme the current implementation only covers%.[prec]e
. Thanks!