Re: [lmi] Using std::chrono::high_resolution_clock for Timer implementation

[Greg Chicares]

On 6/4/21 12:43 PM, Vadim Zeitlin wrote:
[...big snip...] 
>  So, to summarize, I'll confirm that we can rely on getting sufficient
> precision from high_resolution_clock on all platforms we're interested in
> and, if this is indeed the case, will just do the minimal changes to Timer
> class implementation for now

Excellent.

> (and by "now" I mean "after finishing with
> Boost.Regex replacement experimentation"),

Okay. [I'm indifferent to the sequencing of those two steps--whether
high-res-timer precedes or follows regex-experiments doesn't matter
to me: swap them freely if you choose.]

> that won't affect its public
> interface.

I think we agree on a detail that may be implicit, but let me state
it explicitly: I would strongly prefer that std::chrono::* be
physically sequestered as much as possible, ideally by modifying
only 'timer.cpp'. I'd rather not #include <chrono> in 'timer.hpp',
which is included in many other lmi source files.

> Once this is done, we could discuss making the class fully
> inline and, maybe, even some potential enhancements of its API.

Yes, I concur.