[tex-hyphen] new upload to CTAN

[Mojca Miklavec]

2011/1/16 Takayuki YATO (ZR) wrote:
> Hello all,
>
> I investigated which hyphenation pattern Babel would use for each
> language option listed in babel.sty, on the artificial settings
> where every pattern name (including synonyms) listed in language.XXX
> has a distinct \language value (ie. if 'bahasa' and 'indonesian'
> were different patterns then which one would 'bahasa' langauge
> option choose?). I did the test for both language.ptx and
> language.dat lists.
> # NB: The set of "all the hyphenation pattern names", "all the valid
> # language option names of Babel" and "all the names of ldf files
> # provided by Babel" are all-distinct.
>
> The languages for which the results differ is as follows (the full
> list is attached)
>
> Language          language.ptx      language.dat
> australian        australian        british
> canadian          canadian          american
> newzealand        newzealand        british

I see now (from babel):

\LdfInit\CurrentOption{date\CurrentOption}
\ifx\l at english\@undefined
  \ifx\l at UKenglish\@undefined
    \ifx\l at british\@undefined
      \ifx\l at american\@undefined
        \ifx\l at USenglish\@undefined
          \ifx\l at canadian\@undefined
            \ifx\l at australian\@undefined
              \ifx\l at newzealand\@undefined
                \@nopatterns{English}
                \adddialect\l at english0
              \else
                \let\l at english\l at newzealand
              \fi
            \else
              \let\l at english\l at australian
            \fi
          \else
            \let\l at english\l at canadian
          \fi
        \else
          \let\l at english\l at USenglish
        \fi
      \else
        \let\l at english\l at american
      \fi
    \else
      \let\l at english\l at british
    \fi
  \else
    \let\l at english\l at UKenglish
  \fi
\fi

No difference in functionality indeed since there are no special
patterns for Australian. To me this seems like the author of babel was
thinking of: "if someone ever comes up with special Australian
patterns, they will work out-of-the-box".

> As for the languages but last two, the two pattern names
> resulted from the two settings are synonymous in language.ptx.
> As is already argued in this ML, the patttern for language
> "samin" is in fact absent in either setting. The pattern
> "kurmanji" is a real addition and simply favorable.

> kurmanji          english           kurmanji

Kurmanji patterns are new (2009), so I can understand why they are not
present in language.ptx.

> samin             samin             english

However the old setting in pTeX was to use
    samin    sahyph-ptex.tex
which is the equivalent set of patterns as hyph-sv.tex (Swedish). I
hardly know anything about Samin (North Sami), so I have no idea what
would be preferrable.

> magyar            magyar            hungarian

Oh, I see it now. This one was overlooked by me. For babel both would work:

\ifx\l at magyar\@undefined
  \ifx\l at hungarian\@undefined
    \@nopatterns{Magyar}
    \adddialect\l at magyar0
  \else
    \let\l at magyar\l at hungarian
  \fi
\fi

We can make a synonym hungarian=magyar if needed. (The reason against
doing it is that creating a global synonym in TeX Live will load the
largest available patterns twice in eTeX, while it won't really bring
any advantages. I can easily add it back to language.ptx, but that
won't make any difference at all.)

> bahasa            bahasa            indonesian
> indonesian        bahasa            indonesian
> indon             bahasa            indonesian
> bahasai           bahasa            indonesian
> malay             bahasa            indonesian
> meyalu            bahasa            indonesian
> bahasam           bahasa            indonesian

>From what I understand there any difference if we leave just
indonesian? (We can create indonesian=bahasa synonym if needed.)
Anyway, the old "bahasa" was just pointing to Indonesian patterns, so
this change doesn't make any difference at all, I guess.

> As a consequence, there will be nothing bad happening,
> even if a user is using weird language options names
> such as "canadian".

You are right.

A short summary:
- I made some changes in language.ptx to accomodate the changes I made
it TeX Live and hyph-utf8. Please feel free to fix the file in any way
you want (or ask me to fix it).
- List of differences with previous version of language.ptx: some new
patterns added, some patterns updated, samin has been removed (it
pointed to Swedish patterns before), I temporary commented out Greek
(+some different synonyms without any functional difference).
- Let's discuss Greek in the other thread.
- I have another question about dated German patterns for another thread.

Thanks a lot for the very helpful insight,
    Mojca