[XeTeX] "new-babel", was: Ancient Greek hyphenation

[François Charette]

Let me I add again some noise to the discussion...

Will Robertson wrote:
> On the contrary, I think choosing a name early will help us avoid
> calling it "xbabel" without thinking about it.
OK, should we adopt "postbabel" as a working name for now, and vote for 
a more definitive name later? Or should we vote now? The candidates at 
the moment are:
    * tongues
    * multilang
    * polyglotta or
    * polyglossia (used by linguists to denote a situation of 
plurilingualism)
    * meltingpot :-)
    * what else?

------

Concerning the language/script => font switching problem, I fully share 
Jonathan's concerns. But I had also something in mind that resembles 
what Will, Axel and Nicholas have described, namely a setup where fonts, 
font features and other things can be optionally specified for the 
various languages or scripts to be used in the document. In such a 
situation font switching would only occur if the user has explicitly 
asked for it in the preamble. Those ideas are in fact quite similar (if 
I recall correctly) to the implementation in Mem.

A little digression: We should be wary indeed of the fact that not a few 
non-European languages can be written in several scripts, especially 
languages of various Muslim people: Azeri, Kurdish, and several 
Caucasian and Central Asian languages can be (or have historically been) 
written in Latin, Cyrillic and Arabic scripts. Punjabi can be written in 
Arabic (Shahmukhi) or Indic (Gurmukhi) scripts, Malay and many African 
languages in Latin or Arabic. In the past Arabic has also been written 
with Hebrew and Syriac letters by religious minorities. Of course, in 
each case one could circumvent the problem by defining these 
script-related variants as separate languages (calling them e.g. 
kurdishcyr, shahmukhi, jawi, judeoarabic, garshuni, etc.)

Now back to the essential task: Apostolos has just attempted a brief 
summary, and I will leave it at that, as it is already past midnight 
here. But one last thought before I send this: I had a look at babel and 
antomega's code for dealing with hyphenation patterns, the latter being 
a simplification of the former. I think by removing a few lines of code 
in antomega's hyphen.cfg we could easily come to terms with the 
hyphenation issue in our "postbabel" package.

Oh, and I also had a (very quick) new look at mem: AFAICT several of the 
tasks achieved by means of OTPs there are available as font features, 
while most if not all of the remaining ones (such as transliteration) 
could be implemented as TECkit fontmappings. Believe me, TECkit is much 
more powerful than Jonathan would admit! And contrary to OTP it is a 
*very* reliable piece of software :-) (It would be even more powerful 
and flexible if it were possible to process an array of fontmappings 
(concatenation) at runtime, as with Omega. But let's keep that aside for 
the moment: Jonathan faces more urgent tasks!)

Enough noise...

François