[tex-hyphen] hyphenation different in xelatex?

[Claudio Beccari]

I tried to understand the problem and went through the existing German 
pattern files used by (Xe|Lua)Latex and pdfLaTeX.

The behavior that Karl described while starting this thread is hardwired 
in the pattern files themselves; but what astonished me is discovering 
that dehypht.tex is still using special tricks to cope with TeX 2.0. 
Now, independently form the specific patterns, is there any reason to 
maintain any patch to cope with a piece of software that ha been 
superseded more than 25 years ago? Isn't much simpler to filter the 
pattern file for utf-8 aware programs through the utf8 to t1 encoding 
and have the same patterns (coded in a different way) for all engines?

Objection: but vintage files get hyphenated in a different way! OK, but 
the real question is: Is the new hyphenation correct?
If it is, why getting upset? The small differences in typesetting may 
involve a couple of paragraphs in a whole document, and probably no 
reader would notice the difference.

It's like using CM or LM fonts; they are almost identical, but not 
exactly with the same metrics; so there might be some variation in 
occasional paragraphs. Or it's like using or not using microtype: line 
breaks are different with or without this package functionality. Is the 
document typeset in a better way with or without microtype even if the 
result is different from the one obtained more that 25 years ago?

Being not a German I might miss something: being a lover of ancient 
books I might seem to be writing profanities about the fact of getting 
occasional and rare different line breaks in documents printed 25 years 
ago and reprinted today; but, no, I am not: a 25 year old document is 
not an ancient document.

Claudio

On 04/09/2015 01:41, Karl Berry wrote:
> FWIW, I think changing hyphenation in (pdf)latex for any commonly used
> language would cause far more problems that it would help.  Messing
> around with {xe,lua}latex is one thing -- it's reasonable to break
> compatibility there.
>
>      group that their new patterns are not yet stable enough to be used in
>
> Whether the patterns are stable or not, they are different.
>
> Whatever.
>
> k