Things like "heart", "flag", "envelope" or "star" have their unicode. But things like "burger" menu, "share" or "tag" icons don't. What do I do? If I leave unicode empty, then at least Illustrator CS6 treats these glyphs as unknown and won't allow to change their family or weight via the menus.
I know that with San Francisco, Apple utilized Private Use area for some glyphs. Is that the solution? Is there a difference which numbers I use? Can I just start from U+F0000 and go forward?
In theory you could even do more than one of these options! This seems like an especially bad idea, however.
- Use the Private Use Area for characters which are otherwise unencoded. This has the advantage that the glyphs are accessible in just about any app, although the convenience of that access may not be great.
- Make the special characters as unencoded alternates (via some OpenType feature) of characters that are encoded. Then you have to decide which character to make them an alternate of.
- Use some incorrect Unicode and encode them directly, in the wrong slots (like for example A-Z).
- If the entire font is such special characters, you could use the Microsoft "symbol font" approach, which is essentially a modified version of option #1, with some of the convenience of #3.
The last time I made a symbol font it was ~ entirely characters without legitimate codepoints, and intended mostly for non-professionals, so I used option 3. If I was doing those again today I would go with #4.
With approaches 2 and 3, you will get your characters turning into something wrong when the user changes fonts. Possibly with 1 (although quite possibly they’d just get notdefs, depending on fallback). Not sure about 4—that behavior may be dependent on the OS (or the app, in the case of Adobe apps).
So which range of PU do I have to use? Does it even matter?
Those two suggestions stem from the thought that it is better to get notdefs than wrong glyphs. This would be my preference, at least.
Is just putting all icons inside emoji range ok?
Edit: Or this one ☰ 2630. This alchemical symbol might be an acceptable fallback for share. 🜉 1F709 Tag? ⛉ 26C9 ☖ 2616
Quick emoji history: The emo from emoji has nothing to do with emotions. Emoticon is a portmanteau of emotion and icon. E is japanese for picture. Moji is japanese for character/glyph/letter.The original emoji sets from Japanese phone companies didn't have a large percentage of facial expressions. The had a lot of the weather and transportation symbols that still exist in the current emoji set. The first big emoji set (180) came out in 1999 and wasn't that heavy on smileys and had a poo symbol. I think Docomo expanded it to about 1700 and au/SoftBank copied them with their own encodings and a few additions. When the Unicode proposal was made about a decade ago, the emojis submitted were a combination of all the phone companies' custom encoded sets. It's almost identical to that initial SoftBank 1700 set.
In my eyes that means one shouldn't put anything into emoji range to avoid consistency issues.
Huh? Only in text specifically formatted with your font. So, generally, no.