From: Ivan Matek (libbooze_at_[hidden])
Date: 2019-09-24 20:45:09
On Tue, Sep 24, 2019 at 1:51 PM Niall Douglas via Boost <
boost_at_[hidden]> wrote:
> > In the medium-long term, we may be interested in using Boost.Regex to
> replace our regex implementation (when we can break ABI).
>
> You wouldn't prefer Hana's really cool constexpr one instead with the
> much superior performance?
>
> (I appreciate it isn't conforming, but I'd like there to be some day a
> fully conforming implementation using that technology i.e. you to write
> it Stephan will all that free time you have!)
>
>
> I assume the disadvantage of Hana library is that it is not used as much
as boost so it is *probably *more buggy, and I am not sure it can easily be
wrapped to match the std:: stuff since in my experience it is impossible to
dispatch code to constexpr and nonconstexpr implementation.
By this I mean
const std::string s("bla");
std::regex r1(s);
std::regex r2(s + std::to_string(time(NULL))); //NULL to upset STL
Here r1 and r2 look the same to the poor regex constructor, although one is
known at compile time.
disclaimer: I admit I did not follow C++20 standardization so maybe those
nice people from WG21 gave us some magic tools to do this.