← Back to team overview

kicad-developers team mailing list archive

Re: [PATCH 0/4] Performance counter rework

 

Hi jp,

On 06.01.2017 19:43, jp charras wrote:

>>> converting the std::chrono::system_clock::now() to
>>> high_resolution_clock breaks builds on osx, i can't see if there is an
>>> explicit conversion function..... is there any reason we aren't just
>>> using std::chrono::high_resolution_clock::now()? as this builds fine
>>> on osx and i don't really want to have to ifdef it

>> Yes, that is actually correct.

"That" is the suggested replacement.

>> Another case of gcc accepting incorrect code. :/

> I don't really understand your answer.
> What code is correct and what code is incorrect?
> gcc has nothing to do with this issue.

Correct:

std::chrono::time_point<high_resolution_clock> x =
    std::chrono::high_resolution_clock::now();

Incorrect, but acceptable in gcc's implementation:

std::chrono::time_point<high_resolution_clock> x =
    std::chrono::system_clock::now();

LLVM complains about the latter code. MSVC I haven't tested because the
build is currently broken, and ATM I have a customer project that
requires almost all of my time.

   Simon

Attachment: signature.asc
Description: OpenPGP digital signature


References