Werner LEMBERG <[EMAIL PROTECTED]> writes:
>> some basic knowledge still. Below is what I have tried to do.
>
> Thanks for your try.
Perhaps I am trying your patience, but I have more questions,
hopefully you or someone else can answer or point me to more
documentation. I think I am clear on the general idea now, and your
previous explanations went a long way to enabling me to find out the
specifics myself.
>> I took the c42*.fdx files from the JIS subdir of CJK, modified them
>> to become c70*.fdx (minor editing) and put them with the c70*.fd in
>> the UTF8 subdir.
>
> This is not sufficient; see below.
OK, understood that it is not sufficient.
>> I was under the assumption that in the .fdx file the initial
>> \CJKvdef{rotate}{} command would use the graphicx package to do
>> rotation automatically on *all* characters, and that only desired
>> exceptions to this general rule would be given after that command
>> (using \CJKsymbolsimple for instance).
>
> Yes.
>> As such, use of the .fdx file requires the .fd file to be read also,
>> right?
>
> It's the other way round: The CJK package modifies some internal LaTeX
> commands to make a .fdx file be read if available.
Question 1: in vertical.txt and fdxfiles.txt it seems to be stated
that the .fdx file is read *after* the .fd file (and then as you say
modifies some LaTeX-internal commands). Thus, assuming no exceptions
written in the .fdx file, the fact that some glyphs are not rotated
is owing to incomplete subfont mapping? (which I am learning how to
do thanks to your help so far).
Question 2: I do not see a font declaration in the .fdx file so I
assumed the one from the .fd file holds. Or is this some of the
"magic" you refer to below?
>> But I cannot see where the plane and code points are needed in
>> either file unless exceptions are specified in the .fdx file.
>
> This is the very magic done by the CJK package: It hides the plane and
> code point handling completely from the user.
OK. But see my question 2 above.
>> Next, in the file below, I see exceptions written for glyphs in
>> plane "sy", which would be a special name for something in DNP
>> encoding, right?
>
> Yes. The mapping for the DNP subfont layout is given in file
> JISdnp.enc which contains this, for example:
>
> [EMAIL PROTECTED]@[EMAIL PROTECTED]
>
> This means that the first-byte code 0xA1 of JIS encoding is mapped to
> the `sy' font, and that CJK should subtract 160 of the second-byte
> code to get the proper glyph index in that subfont. The `A1'
> parameter has to be used in the [EMAIL PROTECTED]' macro only, to specify
> the used exception list in punctuation tables defined by the
> [EMAIL PROTECTED] and [EMAIL PROTECTED] macros (in file CJK.enc). With other
> words, JIS code 0xA1A1 is mapped to glyph index 1 in subfont `sy'.
OK, many thanks for that explanation. The "TeX Book" was of help here
also, a lot of things became clearer finally, and fear of the macros
is also proportionately reduced.
>> I assume this is what you meant when you said the translation
>> involved "replacing the subfonts and glyph indices with the Unicode
>> pendants".
>
> Yes. The subfonts for Unicode encoded subfonts are always two
> hexadecimal digits (and the glyph indices are different too, of
> course), and all lines in the .fdx files which refer to glyphs must be
> updated accordingly. BTW, it's probably easiest to look up the
> correct mappings between DNP and JIS by checking the file DNP.sfd;
> mappings between JIS and Unicode are given elsewhere. Have a look
> into file makeuniwada.pl which I've written to create Unicode encoded
> subfonts for the Wadalab fonts.
OK, understood. I find I had to use the git repository to get the Perl
script and DNP.sfd, for some reason CJK Version 4.7.0. is not yet on
my Ubuntu Gutsy or Debian unstable, and well, I was hoping to avoid to
error-prone process of installing it myself over the current TeXLive
2007 versions in Ubuntu and Debian.
Question 3: in essence, what I am now faced with is updating/expanding
the unicode subfonts with the help of the files you advised me to
look at.
Thanks,
Gernot
--
Gernot Hassenpflug
_______________________________________________
Cjk maillist - [email protected]
https://lists.ffii.org/mailman/listinfo/cjk
