It is Microsoft specific and is known as a "best fit" mapping, or "best fit
characters." ATAIK it is not documented anywhere, except for being mentioned
(in a slightly negative light!) in a couple of places. Oh, and (just found
this one!) it is mentioned in WideCharToMultiByte:
It mentions the (Win2000 only) WC_NO_BEST_FIT_CHARS, which turns off this
behavior. This would also be slightly slower, I believe, but it is not
doc'ed as such.
There are some amusing mappings, such as the fect that the infinity sign
maps to the number eight as a "best fit" when no infinity is available. And
if you take the Greek alphabet, a few characters (like Gamma will map to G
even though others will not).
I think they did it to make us smile whenever someone would catch one of the
a new book on internationalization in VB at
----- Original Message -----
From: "Adam Twardoch" <firstname.lastname@example.org>
To: "Unicode List" <email@example.com>
Sent: Friday, November 10, 2000 6:12 AM
Subject: National Languages Support in Windows
> When browsing through the documentation of the National Languages Support
> MSDN, I couldn't find one interesting feature which is present in
> When writing text in Outlook Express or Word 2000, I noticed that the
> software converts Unicode text "intelligently" when saving in an encoding
> where some characters are not supported.
> Thus, "aogonek" is being converted to "a", and "threesuperior" gets
> with "3".
> Are those mechanisms part of some public API, or are they Office or
> And, are there some standardized "transliteration" mechanisms for such
> situations, or are those conversions Microsoft-specific?
> Adam Twardoch
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:15 EDT