It seems that Glyphs does not support ‘hist’ feature (when generated automatically). The series of ‘longs’ characters and ligatures are not generated within a ‘hist’ feature, instead they are included in the ‘dlig’ feature (when generated automatically). All historical glyphs should be removed from there manually otherwise the text will appear a bit weird.
Hope it helps.
The longs ligatures in .dlig do not do any harm. If you haven’t some longs in your text, it will not substitute anything. The longs has its own unicode so you do not need a feature to use it.
If you like to use the hist feature, make a copy of the longs and call it s.hist and name your ligatures s.hist_t.
Well, “hist” feature substitutes s by longs.
sub s by longs;
sub longs h by longs_h;
sub longs l by longs_l;
sub longs t by longs_t;
sub longs i by longs_i;
sub longs longs by longs_longs;
For me it is more convenient to gather all these substitutions under the “hist” feature, instead of changing the name of longs to s.hist.
According to http://www.microsoft.com/typography/otspec/features_fj.htm#hist the ligatures should not be in hist, but in liga or dlig.
The case of longs is at least disputed because it does have its own encoding for the case where scripts use both forms of s, as it was the case in written German until mid-20th century.
Then, there is the case where longs is indeed only a form of s with no semantic difference to s. Take Karolingian scripts for example which only ever use the longs shape. In that case, you’d use the s character for storing the text, but may prefer the longs glyph in a layout.
Personally, I generate an identical glyph with longs=s.hist (and do the same for all affected ligatures) and can cover both cases.