Perhaps something very basic. If the old text does not use Unicode encoding, then the ligatures will not work, since the pieces being looked for do not exist, as far as the program knows.

A bit more subtle would be if you've used, say, "theta" for a name, but your statements use the Unicode index, or vice versa.

For example,

sub pi omicron upsilon by pi_omicron_upsilon;

won't work if pi, omicron, or upsilon are *named* uni030c, uni03BF or uni03C5 in your "old text" document. Or even if just one of them is.

For debugging, I'd start assuming something like that, since you're getting scrambled text in English as well. An encoding issue. Remember, FontLab is quite literal, it won't equate a name with a Unicode index within a feature statement, and it won't make the substitution in a text (old or new) if the encoding isn't in the precise form it expects.

Dear Charles
thanks for your answer. Maybe, I have to say that all those texts work properly in InDesign CS4. That's why I'm thinking about an Adobe issue.
Konstantinos Siskakis

I have to say that all those texts work properly in InDesign CS4

Same text files, same font files? We're getting beyond my expertise here, but I'd say, then, it might be language tags. Significant changes in how those interact with CS4 and CS6, I believe. But that shouldn't scramble the English. We don't use language tags in our shop, but of course it is a closed system.

Just to make sure: You have exactly the same text files, and exactly the same font files. Everything is fine using CS4, not fine with CS6. (And you're not using the World Composer with text set as right to left in CS6... No, that'd be noticeable.)

Exactly same text file and same font as you can see. Maybe its something that has to do with language tags. Or a new Adobe issue...
Konstantinos Siskakis

OK. I had misunderstood. I took your original post to mean the ligatures were not working -- that is, were not forming. Everything I wrote was based on that (incorrect) assumption.

For what is actually going on with the two files, someone would likely have to sit down with everything to de-bug. All FontLab will do is let you see the code used for forming the ligatures. Why there is extra space may be due to something in the OT features, but without everything available for examination, one just can't tell. And of course, that may not be the source of the problem.

I would also have to say that ligatures work properly in Illustrator CS6!
Konstantinos Siskakis

I am guessing that maybe the features are not correctly registered to all the writing systems in the font? Perhaps the "old texts" are existing document files that are tagged in the app as being Greek, and thus triggering processing of only the Greek features. The "new texts" may not have any particular language assigned and hence, work fine using default processing.

Forgive me if this is already well-understood and not what is happening here.

Dear Thomas thanks a lot!
I think that you are right!

Excellent, glad to be of help. :)