Writing Code that Works with Different International Formats

WINDOWS   | WINDOWS NT

The information in this article applies to:

• Microsoft Windows Software Development Kit (SDK) for Windows version 3.1
• Microsoft Win32 Software Development Kit (SDK) version 3.5
• Microsoft Win32s version 1.2

SUMMARY

Some European coutries like Germany use a different floating point format that doesn't include the decimal point (.). This has caused problems whtn the actual numbers use the decimal point format (xxx.xx) and the German version of Microsoft Excel is expecting a comma (xxx,xx).

Some developers have made the mistake of hard-coding formats or searching for all periods and replacing them with commas. This is not good programming practice.

Microsoft Excel uses the Control Panel Number Format for its separators. So, regardless of what language version of Windows it runs on, it always uses the correct separators. Microsoft recommends that you use the same approach. Always use the separators defined in the number format of control panel.

MORE INFORMATION

All the international information is stored in the WIN.INI file under the [intl] heading. The application should look for the value of "sDecimal." It can query this value using the GetProfileString API. If the application uses this approach, it will have the correct separator for all countries.

To avoid potential problems, never make assumptions about number formats, currency formats (that is, don't hard-code the dollar symbol), date formats, or time formats. These are different for different countries. Use the settings from the Control Panel.

If an application does any text processing, such as sorting or upper/lowercase conversions, it should use the Windows APIs and not supply its own conversion tables and functions. For example, if an application use APIs such as AnsiLower, it will work regardless of language because it obtains the data from the language DLL.

To output floating point numbers in Windows, Microsoft recommends that you use wsprintf. The concept of an "international format string" only has meaning when the string is output to the user. Internal to the computer, for all calculations and manipulations, the concept does not apply. Thus the application can keep its floats in the native format and only convert to strings before calling the wsprintf function.

There is no need to load conversions from float to string, and vice versa. In other words, there is no need to write any functions such as InterStringToFloat(), InterFloatToString(), and so on.

Here's a simple algorithm you could use.

1. Convert your float to a string by using the _fcvt or _gcvt standard C functions.

2. Get the sThousand and sDecimal separators from the WIN.INI file by using the GetProfileString API.

3. Call wsprintf to output the converted string, using sThousand and sDecimal in the formatting string.

   if language=English
      Output_English_Float ()
   else
      Output_International_Float ()

Also, for ease of localization, a developer should place all strings in the resource file. This avoids having to search through all the C files looking for strings to translate.