Most of what you've listed are stylistic variants, and are not assigned official Unicode points. Adobe has used Unicode PUA ranges in the past for some stylistic variants, but has reportedly deprecated this practice.

There are some documented agreements on character assignments in the Private Use Area, notably Medieval Unicode Font Initiative and ConScript Unicode Registry. There may be others but I am not familiar with them. You may find some of the glyphs you want to encode have points within those ranges, and you can avoid thos ranges if there isn't a clear fit. The PUA is a big space.

In the Adobe Fonts the unicode value assigned to small caps or swash capitals for example is not the same in all the fonts.
It seems like such glyphs are left for Open Type features to handle, but then what's the point in assigning unicodes? Is it good practice that every glyph has a unicode value?

I'm sorry if my questions are slightly off subject but I didn't want to start a new thread with a very similar subject.

Some common applications have been reported to not display a glyph if it does not have a Unicode number. This is bad practice on the software publisher's part. Does anyone remember with which programs this is the case, or with those where it has been resolved, with what version?

OpenType features ought to be sufficient to handle such stylistic variants.

Some common applications have been reported to not display a glyph if it does not have a Unicode number. This is bad practice on the software publisher's part.

Supporting Unicode is different than supporting OpenType. A program that did not support Unicode these days would be the equivalent of a program not "supporting" ASCII in days past. I suppose you could think of the PUA of Unicode being equivalent to high-order ASCII.

The point of a Private Use Area is that it has no defined use. Asking for an agreement of typographic niceties shows a certain insensitivity to the purpose of Unicode. Consider documents that pass from one use to another, one font to another. That is job of Unicode.

It is a little point, but I abhor fonts with titling caps built into the font. I deal with large amounts of running text (books). Titling glyphs are almost always used for display, and with most work, display usually involves a separate font. If titling glyphs are built into a font, and an application program such as InDesign can't handle the switch to them easily, through a paragraph style, they are not as useful as they would be in a separate font.

As is so often the case, people who use type often have different ideas than people who design type.

My general advice is not to use Private Use Area codepoints for variants of standard encoded glyphs. The PUA should be used only when you have some idiosyncratic symbol that is neither a standard encoded character nor a variant form of such a character. An obvious example would be a corporate logo in a font. The only time I use PUA characters is for decorative, ornament glyphs.