This should happen only with fonts that have internal non-Unicode legacy 8-bit mappings (for Windows, or national asian standards like JIS, EUC, HKCS...). But not for the Unicode mapping of the same fonts.
The UCS mapping of these fonts should only map true backslash on U+00, and map the respective currency sign with its Unicode code point. Then the renderers will use the correct mapping in the correct context (of the encoded text). Web browsers will simply ignore the legcy mappings, and will internally convert the encoded web pages encoded with legacy codepages to the UCS, and render them only with the UCS mapping of the same fonts. Fonts that use "fake" mappings for the backslash in their UCS map are clearly not compliant. However fonts that *only* contain a mapping for the legacy 8-bit codepage are compliant even if they replace their backslash with an asian currency sign (web browsers may attempt to use these fonts even if there's an ambiguity about how to remap their internal legacy encoding from the UCS). These non-UCS fonts still exist (many of them used as core system fonts) but should be deprecated, or upgraded with a version that include the additional UCS mapping (or only this UCS mapping which is sufficient even for use with texts encoded with legacy codepages). However it is not always easy to upgrade these fonts if they are part of the core system fonts (they will be upgraded only when upgrading the OS itself, if it's possible on the device firmware, or if it does not break existing applications depending on the legacy OS version...) 2013/10/23 Denis Jacquerye <moy...@gmail.com> > Michael S. Kaplan had a blog post about this in 2005: > > http://blogs.msdn.com/b/michkap/archive/2005/09/17/469941.aspx?Redirected=true > > In Japanese 円 is generally used for yen, in Korean 원 is generally used > for won, but both the yen sign and won sign are also used. > Some somewhat-Unicode Japanese or Korean fonts have Yen sign or Won > sign respectively instead of backslash in part due to what Kaplan > explains. > Obviously things get messy when copying the backslash character from a > Japanese font to a Korean font if that character has a different > currency glyph in both fonts. > > >