Reading through my previous post I realize that the following link might also be helpful.
As I understand it, the renderer of a wordprocessing package which can handle glyph substitution for ligatures works, for each character in turn, down the table in relation to that character and the next few characters after it, stopping when it gets a substitution or when it gets to the end of the list and finds that no glyph substitution is needed for that character. Thus, in relation to the ligatures for ffi and ffl, I have placed them before ff in the list.
Suppose that the word office is being considered for glyph substitution. Firstly the system looks for ligatures starting with the letter o and finds none. The system then starts looking for ligatures starting with the letter f, finds ffi and realizes that this is a match for the current and next few characters of ffice, so uses that and then proceeds to look for ligatures starting with c, finds ct and realizes that this is not a match for ce (ce being from the remnants of office after o and ffi have been used up) and so does not use it, just using c.
I think that the system would then try to process e, even though it is at the end of the word because I think that it is possible, though I am not sure on the matter, to have a ligature of e and a space so as to have a swash e at the end of a word if one chooses to design a font in that way.
However, had the font had the ligature rule for ff before the ligature rule for ffi, when the system started looking for ligatures starting with the letter f, it would have found finds ff and would have realized that it was a match for the current and next character of ffice, so would have used that and then proceeded to look for ligatures starting with i and then having found none, then proceeded to look for ligatures starting with c and so on as before.
Now, as it happens, Chronicle Text has a number of these ordering matters regarding ligatures.
There is the cha che cho ch collection.
There is the ppe pp pe collection.
There are various long s long s something ligatures as well, and also various other ff something ligatures.
Incidentally, I found some of those ligatures in a pdf in the http://www.waldenfont.com/ webspace. On the German Fraktur Blackletter Fonts at the Walden Font Co. web page there is a manual available in the Downloads section, http://www.waldenfont.com/downloads/gbpmanual.pdf and there is a table on page 14 of that manual.
Then there are my hypothetical metal type simulation ligatures, such as bi and so on, which I mentioned in the following thread.
It might seem unnecessary to include them as ligature substitutions and it probably is unnecessary but I am probably likely to do it anyway as a sort of aesthetic completeness for the font.
Another interesting point is that the experimental font also still allows access to every ligature glyph using a Private Use Area codepoint: I am hoping to produce a fully working OpenType font with automatic glyph substitution yet I am keeping the Private Use Area codepoints as well.
William Overington
18 March 2007
Please note that some of the original text of this post has been deleted by the author and a later post produced on 19 March 2007: that later post is in this thread.