Why is that the same unicode character, in the same machine, can has two (or maybe more) different designs? Check this screenshot: http://i38.tinypic.com/2pqqntj.jpg
Unicode specifies a "character", which is somewhat an abstract entity and is different from "glyph", which is a visual representation of the character. One character can be mapped to different glyphs. Usually glyphs are determined by fontset. In your case I assume glyphs of the snowman character in the fontset for the text entry and in the fontset for the text display are different.
It is a common experience for East Asians; the same CJK kanji character sometimes has different glyphs for Chinese and Japanese. It is still quite a hassle to fix all fontset to show them right (I gave up.)
I don't know a lot about unicode, does anyone know if there is a solution to the "I'm doing the obvious wrong thing by rendering an incorrect glyph" problem? It seems like it would be pretty cool to have rendering engines report back code points that they were unable to render, allowing the application to signal to the user that something is wrong, or even better yet, offer to launch a service to find a font that has that code point.
we're pre alpha testing a practical application of unicode if anyone's interested.. the tool uses intertubes to help ourselves learn more language.. it'll get better when we learn how to incorporate IDN.. no, we're not that smart yet, but yes, we are open source :)
http://www.♀.com/
http://www.☯.com/
http://www.☺.com/
http://www.☼.com/
http://www.♫.com/
Just some from my bookmarks...