zoned_time

[LukashonakV]

Hi @vitaut , this PR provides zoned_time formatter.

---

#include <fmt/chrono.h>  
#include <spdlog/spdlog.h>  

using namespace std::chrono;  

int main(int argc, char* argv[]) {
  const std::chrono::zoned_time now{locate_zone("America/Chicago"),
                       floor<seconds>(system_clock::now())};
  std::string fmt_result{fmt::format("{:%H:%M %Z %z}", now)};
  std::string std_result{std::format("{:%H:%M %Z %z}", now)};

  spdlog::info("\nfmt_result: {}\nstd_result: {}", fmt_result, std_result);
  return 0;
}

With the result:

[2026-01-09 02:09:56.643] [info] 
fmt_result: 17:09 CST -0600
std_result: 17:09 CST -0600

[vitaut]

The formatter should be conditionally defined since zoned_time is a C++20 feature.

[LukashonakV]

Hi @vitaut , yes I've checked CI logs and is trying to solve it

[LukashonakV]

Now it should be fine.

[LukashonakV]

Hi @vitaut , can you please review once again, when you have a time of course. Thanks)

[vitaut]

Thanks for the PR.

[vitaut]

Overall looks good but please add a unit test to chrono-test.cc.

[LukashonakV]

Overall looks good but please add a unit test to chrono-test.cc.
Sure, let me do it today. Thanks

[LukashonakV]

Hi @vitaut , tests are added. Hope they are ok.

[vitaut]

Looks like there are some CI failures.

[LukashonakV]

Looks like there are some CI failures.

Yeah, had not time to check . Will try propably tomorrow

[LukashonakV]

Looks like there are some CI failures.

Yeah, had not time to check . Will try propably tomorrow

[vitaut]

I think we should only provide this specialization if tm supports time zones. Otherwise the timezone information will be ignored. This is basically why the tests are failing.

[vitaut]

@LukashonakV do you still plan to update this PR?

[LukashonakV]

Hi @vitaut , sure. Apologies for the delay. A bit busy with main activities. Hope next week will be able to check. Actually I saw the problem and was wondering about the solution..

[vitaut]

No hurry, just wanted to check if we should keep this PR open. Since timezone names are system-specific I think it would be fine to limit tests to Linux (and we probably don't need to test so many of them, one or two should be sufficient).