In my work I require many characters that the average user never looks at, like alternate ampersands, long s, etc.
Short answer: You are not asking about characters, you are asking about glyphs, and the glyphs you are asking about are not going to be searchable in your PDF, if you plan to publish it online. Depending on what you are doing, anywhere from 5% to 50% or more of your composition will be completely unsearchable.
Long answer: In the technical terminology of ISO10646/Unicode/TrueType, alternates are NOT characters, alternates are glyphs. Glyph alternates are, therefore, not in the same human interface as ISO10646/Unicode characters, but in a separate human interface. For the human interface in Apple Mac OS using QuickDraw GX, see the Typography Palette for the set of feature selectors that are implemented in the individual intelligent font; for Mac OS 9, 9.1 and 9.2, see WorldScript which is a separate install from the retail system software CD. For the human interface in Apple Mac OS X using ATSUI Apple Type Services for Unicode Imaging and CoreText, see the Apple Typography Palette for the individual font, that is, first pick the font in the Font Palette and, if the font has support for feature selectors that draw glyph alternates, then click the Advanced icon at the bottom of the Font Palette to open the Typography Palette for that font. Microsoft implemented TrueType in 1992, but withou support for glyph alternates. Microsoft has implemented support for glyph alternates for Windows Word 2010 (
http://office.microsoft.com/en-us/word-help/opentype-options-in-the-font-dialog- box-HA101809106.aspx).
Per the Unicode Proposal of 1988, per the TrueType Specification of 1990, and per ISO-IEC Technical Report 15285, a distinction is made between imageable glyphs that can be drawn by direct depiction onto characters in ISO10646/Unicode (we can call these preferred glyph forms) and imageable glyphs that cannot be so drawn (we can call these presentation glyph forms). We do not want to encode presentation forms in ISO10646/Unicode, for instance, a different character code for each and every different glyph alterntate for the et ligature of which there are very, very, very many - cf Jan Tschichold, "Formenwandlungen der Er-Zeichen", Frankfurt am Main: D. Stempel AG. Stylistic alternates are allowed in typography, but stylistic alternates are semantic allographs, and semantic allographs are not allowed in telegraphy (ISO10646 is heir to ISO646 and ISO646 is heir to a century of International Telegraph Alphabets back to IT11:1874 which introduced the Baudot character model, see the Unicode Proposal of 1988 for further information).
In the intelligent font model, presentation forms are not drawn by direct depiction onto character codes in ISO10646/Unicode. Mapping from public and font-independent character codes to private and font-dependent glyph codes is done in the CMAP Character Map of the intelligent font model. This mapping yields the preferred or default glyph form as drawn by the type designer. To draw presentation forms, the font-dependent glyph codes output by the CMAP are remapped to other font-dependent glyph codes using human-sensible feature selectors, that is, meaningful typesetting terms. Apple's supplements to the TrueType Specification use the MORX Metamorphosis Extended tables to draw glyph alternates; Microsoft's supplements to the TrueType Specification use the GSUB Glyph Substitution tables to draw glyph alternates. Neither composition drawn by Apple MORX tables nor composition drawn by Microsoft GSUB tables is searchable in Apple's system PDF service.
/hh
