`Bernard GAULLE' wrote
>   >>>>> On Wed, 17 Nov 1999 15:27:12 +0100, "Denis B. Roegel" <[log in to unmask]> said:
>   >   DR> To some extent the same goes with encodings, both input and output.
>   >   DR> Given the features of TeX, the choice of the output encoding
>   >   DR> is relevant for hyphenation,
>   >
>   >   hum, i guess you wanted to say: hyphenation is relevant of the font in use.
>   DR> No, I meant what I wrote.
>   well that wording is inappropriate. We should say: hyphenation depends
>   of the font in use.

I maintain my wording.

>   I wrote:
>   >   no, any TeX using mltex option is independent of encodings. You can
>   >   use CM, EC, Times, ... what you want with any related encoding.
>   >   It would be so easy if all formats around the world were done
>   >   with the mltex option. It's free, standard (as an option can be) and
>   >   without any danger. But life is different ;=)
>   DR> I meant "the brand of TeX you are using is *usually* not
>   DR> independent of the (font) encoding you use."
>   it should. Only few TeX engines don't accept all font encodings,
>   which ones?

It is not what I meant. I meant that there is usually a relationship
between the brand of TeX one uses and the font encoding one uses
(in a given language context). I never said that most TeX brands do not
accept all font encodings.

Of course, there is again a problem of definition. What exactly
is a brand of TeX? I think of a brand of TeX as what you get when
you start creating the format. For me, tex and tex -mltex are two
different brands. Even with web2c7, I am convinced most people
do not use an mltex format. It doesn't mean they will not use
it in the future, I just think that currently it is not the case.
If it were the case, it would not have been necessary to write
a paper on that subject in the "Cahiers GUTenberg."

>   DR> Who uses mltex?
>   DR> Some people, but not everybody.
>   so you should come back to Word: everybody use it...
>   In other words i prefer to have plus than minus, but everybody
>   is free to use what he wants.

The problem is that what seems to be a plus to you has been at times
a minus for others. The reason for this is very simple: it is a matter
of support. Too few people use MlTeX, so too few people can help you
when you have a problem. Of course, I think things have changed, in that
MlTeX is now more solid than it used to be, but a few years ago,
I switched from MlTeX to TeX exactly because I had a terribly hard time
getting help for my problem. And I got no help from GUTenberg
who promoted MlTeX. I noticed that some fonts had been kludged
in order to circumvent a bug in MlTeX (which has been corrected since).
The knowledgeable people from the TeX community were not knowledgeable
in MlTeX (except for Bernd Raichle, of course).

The comparison with MS Word is not appropriate, because one of
the reasons I would not use MS Word now is the same reason
I have not been using MlTeX in the past:
lack of *appropriate* support. (There may be
support for MS Word, but not appropriate for me.)

>   DR> Besides, I haven't followed the latest developments of MlTeX,
>   DR> but it used to be the case that fonts had to have a certain
>   DR> number of free slots for the "fictitious characters" MlTeX
>   DR> creates, in case these \charsubdef is used.
>   no, they are not "reserved" but "used" e.g. any new glyph defined
>   by \charsubdef is associated to a slot in the font, as any other
>   glyph...

Did I say "reserved" ?
There used to be a constraint that you have to use empty slots.
I do not know the current state of things. I know only that
one constraint has been lifted, and this was a constraint that
caused me the aforementionned trouble, namely that only a continuous
tail segment of free slots could be used. Since some PostScript fonts
had used a slot in the middle of emptyness, this spoiled a whole
range of empty slots and crippled MlTeX at some point. But this
problem has been solved. Question: is there an example where MlTeX
is used for remapping 8 bit fonts into (differently encoded) 8 bit fonts?
(Currently, I know for instance that MlTeX can remap 7 bit OT1 fonts
into various 8 bits encodings, but can it do 8 -> 8 ?)
If not, then MlTeX still has its limitations with free slots,
and not every font encoding can be used with it (and remapped).

>   DR> Is this no longer true? If it is still true, it means
>   DR> that some fonts cannot be used in MlTeX and at the same time
>   DR> remapping/virtualization to occur.
>   no, at one time one use 1 input encoding, 1 output encoding and 1 font.
>   Input encoding let TeX know which glyph is targeted.
>   Output encoding let TeX associate a slot in the font which is related
>   to the tageted glyph.
>   When using MlTeX option you only say to TeX to pick up 2 slots of the font to
>   produce the glyph inside the dvi (and let on the side the original font slot).
>   So, encodings should never differ and the only scholar-case for which
>   a pb could occur is when the 2 basic slots (in case of e-acute, the "e"
>   and the "acute" slots) are different in the fonts in use. In nearly 14 years
>   i use MlTeX i never fell on this pb, but you certainly did...

No, I didn't fall on this problem, but the problem I had was both
related to the bugs in MlTeX and to kludged PostScript fonts.
The fonts had been modified to work with the bugs of MlTeX.
And at some point, I used *correct* fonts, and then everything
broke. Of course, some of this happened when I started using
LaTeX2e (I guess I must have been the first person in France
to download LaTeX2e in December 1993), hence I was in a non-GUTenberg
supported configuration. It took me months to figure out what
was wrong.

Denis Roegel