>>>>> "Asger" == Asger K Alstrup Nielsen <[EMAIL PROTECTED]> writes:
>> >>>>> "Lars" == Lars Gullik Bj�nnes <[EMAIL PROTECTED]> writes:
>>
Lars> /** We could have a Uni2T1 filter so that we could ditch the
Lars> inputenc package. This would give use direct access to all
Lars> characters in the T1 encoding. That would be nice! */
>> What do you mean? T1 is a font encoding only relevant for output
>> font. You cannot write a LaTeX document in T1 encoding as far as I
>> know... I do not think there will be any way to avoid using an
>> input encoding package (except of course if we output plain 7bit
>> text).
Asger> What encoding is used in a raw LaTeX file without any input
Asger> encoding packages involved? What encoding is used when we do
Asger> use an encoding package?
Without an encoding package, what TeX accepts is
- plain 7bit for plain old TeX
- iso-latin1 for MlTeX (might be configurable for others)
- Unicode (I think) for Omega
Asger> When I am talking about an encoding, I speak in a general
Asger> sense:
Asger> Accented character constructs can be considered one character,
Asger> just with a multi-width encoding.
Yes, a TeX construct is just another multi-byte encoding, and they
should be considered at the same level.
Asger> The plan I have in mind in connection with the Encoding class
Asger> approach I suggest, is to implement a hierarchy of LaTeX
Asger> Encodings, matching each input encoding package there is (and
Asger> the one without any).
We could setup some (maybe LaTeX) script which reads the input
encoding files and spits out C++ or plain text files usable for
LyX. This way, we let the burden of ensuring these are right to
others.
Asger> "EncLaTeX" will produce plain 7 bit text, where all accented
Asger> characters are done with the LaTeX commands.
Asger> "EncLaTeX_ISO8859_1" (or maybe another name) will do the same,
Asger> except that the characters that are supported in ISO8859-1 will
Asger> be spit out as raw text. EncLaTeX_ISO8859_1 will probably be
Asger> derived from EncLaTeX. Maybe even with multi-inheritance to
Asger> derive from EncLaTeX and EncLatin1 simultaneously. Whatever is
Asger> easiest.
Right.
Asger> To implement this, I need help with understanding the different
Asger> encodings that apply in LaTeX. Is there a reference somewhere?
You can have a look in your distrib to fntguide.dvi for font encodings
and to inputenc.dvi for a (short) description of input
encoding. Basically, all that an input encoding package does is
translate the input encoding to plain old TeX. Reading files like
latin1.def is also probably a good idea.
JMarc
