Hi Georg, thank you very much for this insight. I really like the EB Garamond. It's a wonderful typeface for reading. I used it to typeset books with more than 500 pages together with lualatex and the selnolig package which produced good results with no effort. I was disappointed to learn that there is no package like selnolig for ConTeXt but the book I am now typesetting is only 120 pages long and I'll switch ligatures off manually with the ZWNJ replacement. (The ConTeXt notation is a bit uncomfortable, but as I only have to use it once...) I only found one issue with ZWNJ. It splits up the words so that they are not found by the search. If ZWNJ is the right way to do it, I'll start a tiny project on Github with a replacement file. First time I hear about the new keyboard layout. I think I want my T2 keyboard now! ;-) juh Am 18.01.2016 um 23:38 schrieb Georg Duffner:

Hi Jan, hi Hans,

This is the developer of EB Garamond. I’m only now and then reading this list so I only learnt about your troubles now.

The way the ligatures are composed simplifies the creation of more ligatures after the same scheme while not blowing up the font by having to create one glyph per ligature combination but only one per component. In fact, it’s not just kerning, like Hans said, but one-by-one glyph substitutions. It might be seen as inconsistency that one feature can be executed in different ways. I know of that problem but it’s not that inconsistent at user level if you use the methods intended by unicode and opentype to deal with breaking up ligatures. The symbol ZWNJ (zero width non joiner) is exactly meant to do this so if you insert it, you should get what you want. What’s more, by inserting ZWNJ you should also be able to benefit from kerning between the non ligated f and the following letter or from contextual alternate forms that fit better than the long bowed f, if you want that. With EB Garamond, this should be true for both, engines that handle ZWNJ correctly or not.

By the way, the newly standardized German keyboard layout even has the ZWNJ on it and the norm recommends its use to prevent ligation.

Best regards Georg Duffner

Am 18.01.2016 um 11:09 schrieb Hans Hagen:

...

On 1/18/2016 10:16 AM, Jan U. Hasecke wrote:

...

Am 17.01.2016 um 20:05 schrieb Hans Hagen:

...

On 1/16/2016 3:58 PM, Jan U. Hasecke wrote:

...

Am 16.01.2016 um 13:31 schrieb Schmitz Thomas A.:

...

Please provide a minimal example of your problem. It’s impossible to help when we have no clue what you’re doing.

Sorry, of course.

After setting up a mwe I found that it is a font related issue.

When I don't specify a font, it works. --> example.tex

When I choose EB Garamond, it does not work. -- example-Garamond.tex

I confirmed this behaviour in my real setup.

don't assume that ligatures are always real ligatures ... in that font it's just kerning .. this kind of works okay:

\replaceword[sellig][auflösen][auf{-}{}{\zwnj}lösen]

I am confused as the specimen of EB Garamond mentions (real) ligatures. They are listed as glyphs.

https://github.com/georgd/EB-Garamond/blob/master/specimen/Specimen.pdf

maybe the archaic st ligature is a precomposed but f f l i aren't done that way but by either kerning or replacement of individual glyphs + kerning (there are many methods for this) ... also, 'liga' might mean ligature but in practice is used for all kind of things ... in opentype 'ligature substitution' is just a many-to-one replacement but that doesn't mean that 'liga' uses that ... welcome to the inconsistent open type mess

Mh, yes. :-(

Two additional questions. Shall I file a bugreport for this issue? What would be the right words: please provide real ligature glyphs instead of composed ones?

EB Garamond is a free font also in the sense free of charge. But what can I expect when I buy a commercial font? I would be quite annoyed when I buy a font which does not provide the features in a way that I can use them in ConTeXt.

Is there a font quality page on the Wiki with a feature comparison?

juh