Allow customisation of padding character for chrono

[bradrn]

Reading through the chrono documentation, it appears there is no easy way to customise the padding/prefix character for presentation types like C, d, e, g, H, I, etc. Thus, it seems impossible for me to format a time string like ‘1:08 PM’: the closest I can get is ‘01:08 PM’. It would be nice to have a way to customise the padding character… perhaps the modifier field could be extended so that characters other than E or O are treated as the character to use?

[vitaut]

Thanks for the suggestion but these format specs mostly follow strftime / standard chrono specs and there are no plans to significantly deviate from them.

[bradrn]

Ah, didn’t realise there was a standard around these things. Thanks for letting me know! What a pity that C strftime doesn’t support this.

That being said, one Python resource suggests that %-I might do what I want — and indeed, Python strftime("%-I") seems to do what I want when I test it. I know this format code is nonstandard, but it seems like quite a useful thing to have. (In my particular case, I’m using fmt via Waybar, and it would be nice to display the current time in a way which I find most readable.)

[vitaut]

- has been implemented in #3271 since it's a common strftime extension.

[bradrn]

In that case, there must be a bug, because this program:

#include <fmt/chrono.h>

int main() {
    auto t = tm();
    t.tm_year = 2010 - 1900;
    t.tm_mon = 7;
    t.tm_mday = 4;
    t.tm_hour = 1;
    t.tm_min = 15;
    t.tm_sec = 58;
    fmt::print("{:%-I}", t);
    return 0;
}

Fails with the following error message:

terminate called after throwing an instance of 'fmt::v9::format_error'
  what():  invalid format
Aborted (core dumped)

Whereas the same program works fine when I use %I instead.

[vitaut]

Works on godbolt: https://godbolt.org/z/3Gona4K4P. You might be using an older version.

[bradrn]

You might be using an older version.

A-ha, so I am!

$ pacman -Q fmt
fmt 9.1.0-4

And indeed, when I try that version on Godbolt, it gives that error. I wonder why the Arch package is lagging behind… in any case, apologies for cluttering up the repo with this issue!