I have a couple of issues with std.string.format.

1. why
writefln(format("%%1.%df",2));
produces 'Error: std.format'
but
writefln(format(format("%%1.%df",2),1.0));
works fine?

2. on Windows:
format(format("%%1.%df",4),666666/1000000.0) produces "0.6667"
but
format(format("%%1.%df",4),66666/1000000.0) produces "0.0666"

on Linux second line gives correct answer "0.0667"

What can I do with the second problem? Can anyone suggest a workaround?

Egor Starostin wrote:
> I have a couple of issues with std.string.format.
> 
> 1. why
> writefln(format("%%1.%df",2));
> produces 'Error: std.format'
> but
> writefln(format(format("%%1.%df",2),1.0));
> works fine?
> 
> 2. on Windows:
> format(format("%%1.%df",4),666666/1000000.0) produces "0.6667"
> but
> format(format("%%1.%df",4),66666/1000000.0) produces "0.0666"
> 
> on Linux second line gives correct answer "0.0667"

This sounds like a C library issue. Are you using GDC, or DMD ?
> 
> What can I do with the second problem? Can anyone suggest a
> workaround?

Egor Starostin wrote:
> I have a couple of issues with std.string.format.
> 
> 1. why
> writefln(format("%%1.%df",2));
> produces 'Error: std.format'

Because the format function returns "%1.2f", so writefln expects an extra floating-point argument after that string.
What you may be after is
	writefln("%s", format("%%1.%df",2));

> but
> writefln(format(format("%%1.%df",2),1.0));
> works fine?

Here the outer format call returns "1.00" which is a regular string, so writefln() doesn't have a problem with it.


If you're passing a string to writefln() that contains % signs you want to be sent to the console, either make sure they're properly quoted (i.e. %% instead of %) or pass a format string that expects a string before it (e.g. "%s").
If you don't want them to be sent to the console but want them to be interpreted and substituted, make sure you supply the correct number (and type) of arguments...

> > 2. on Windows:
> > format(format("%%1.%df",4),666666/1000000.0) produces "0.6667"
> > but
> > format(format("%%1.%df",4),66666/1000000.0) produces "0.0666"
> >
> > on Linux second line gives correct answer "0.0667"
> This sounds like a C library issue. Are you using GDC, or DMD ?
DMD

In Python for Windows this code works correct. Does this means that Python doesn't use C library in such case?

> > 1. why
> > writefln(format("%%1.%df",2));
> > produces 'Error: std.format'
> Because the format function returns "%1.2f", so writefln expects an extra floating-point argument after that string.
Indeed! Thank you for an explanation.

Egor Starostin wrote:
>>> 2. on Windows:
>>> format(format("%%1.%df",4),666666/1000000.0) produces "0.6667"
>>> but
>>> format(format("%%1.%df",4),66666/1000000.0) produces "0.0666"
>>>
>>> on Linux second line gives correct answer "0.0667"
>> This sounds like a C library issue. Are you using GDC, or DMD ?
> DMD
> 
> In Python for Windows this code works correct. Does this means that Python doesn't
> use C library in such case?
It uses a different C library (the one which Python was compiled with). The one DMD uses was written by Walter, and supports 80-bit floats, for example.