formatting double with fixed format, but no trailing zeros about fmt HOT 9 CLOSED

Looks the same as #2155.

from fmt.

I don't think you can express it in a format string directly but you can use the f specifier and then remove trailing zeros and the decimal point.

[As you meant: #2155]((#2155 (comment))
Of course it works, even with printf - I would just like it to be more convenient, as would others.

The format string syntax is pretty much fixed at this point for compatibility with std::format. That said you can implement this functionality via the extension API without changes to the core library.

that it is close to the standard - is very good, but why not extend it, that does not harm the fmt library, if the standard works, but there would still be extensions, everything so far continues to work exactly....

from fmt.

Hi,

i implemented the new type in the current version. Do you want to see the code?

this is an example output for the double value:

	double w = -0.00000001;

	fmt::print("{{:+.3v}}={:+.3v} {{:+#.3v}}={:+#.3v}\n", w, w);
	fmt::print("{{:-.3v}}={:-.3v} {{:-#.3v}}={:-#.3v}\n", w, w);
	fmt::print("{{:.3v}}={:.3v} {{:#.3v}}={:#.3v}\n", w, w);
	fmt::print("{{:*>12.3v}}={:*>12.3v} {{:#.3v}}={:#.3v}\n", w, w);

output:

{:+.3v}=-0 {:+#.3v}=-0.
{:-.3v}=-0 {:-#.3v}=-0.
{:.3v}=0 {:#.3v}=0.
{:*>12.3v}=***********0 {:#.3v}=0.

from fmt.

should i make a PR to be able to present the enhancements?

from fmt.

No but you can post an example of a user-defined formatter that does this.

from fmt.

As i understand it, i can't define a user-defined formatter for a standard type like double. that's why i integrated the new type for formatting. i only have to add/change about 40 lines. everything existing should work as before.
is there a change that this change can be integrated into the library in an upcoming version?

If not, what are the reasons for this. I would be happy to start a discussion on this.

from fmt.

You can define a custom formatter for this via a wrapper type.

from fmt.

I apologise for being so harsh, I don't want to annoy anyone.
I have understood that and I know how to implement it.
What are the reasons for not including the new format type?

• performance?
• no standard?
• no need?
• no time/resources to implement?

in my use case i sometimes have more than a million double values to format. just to wrap the value in a wrapper type and then edit it in a custom formatter makes no sense and costs unnecessary performance.

the use case is quite common in industry, especially with nc machines, such as milling and turning machines, also 3d printers. it is nc code or gcode. the new format reduces the file size and is more readable.

from fmt.

It is important to maintain compatibility with std::format and keep the core library small. The extension API is there to support custom, possibly common, formats.

from fmt.