xetext-pdftex differences

xetext-pdftex differences

[Steffen Wolfrum]

Hi,

as there are some on this list working/playing with xetex/context:
Where do I go when I experience differences between context/pdftext and context/xetex?

For example with context/xetex I get ...

! Undefined control sequence.
\stopframedtext ->\dostopframedtext 

... and I do get "SS" instead of eszett ("ß") with the following:

\usetypescriptfile[type-gyr]
\usetypescript[times][ec]
\mainlanguage[de]
\enableregime[utf]
\setupbodyfont[times,9pt]

Hello Thermes! 1234567890 üööäß


Whom to tell this?

Steffen


P.S. Besides these details it is amazing how smooth ConTeXt files run with XeTeX!!

Re: xetext-pdftex differences

[Mojca Miklavec]

On 4/8/07, Steffen Wolfrum wrote:
> Hi,
>
> as there are some on this list working/playing with xetex/context:
> Where do I go when I experience differences between context/pdftext and context/xetex?

What do you mean?

> For example with context/xetex I get ...
>
> ! Undefined control sequence.
> \stopframedtext ->\dostopframedtext

Which version of ConTeXt? It works OK here with 2007.03.22, but I need
to try the new beta. In principle this should not happen, it's
probably a bug or some other local problem.

> ... and I do get "SS" instead of eszett ("ß") with the following:

Interesting effect ... (as if the encoding was unknown)

> \usetypescriptfile[type-gyr]
> \usetypescript[times][ec]

This should be
    \usetypescript[times][uc]
but in principle it shouldn't work without additional definitions in
typescript files. See below.

> \mainlanguage[de]
> \enableregime[utf]
> \setupbodyfont[times,9pt]

Or, to prevent from an additional source of problems (overloading
times properly):
    \definetypeface[times][rm][serif][termes][default][encoding=uc]
    \setupbodyfont[times,9pt]

> Hello Thermes! 1234567890 üööäß
>
>
> Whom to tell this?

Mailing list is the perfect place.

OK, so here are some additional definitions (which should probably go
to type-gyre.tex one day):

A complete list of synonyms:

\definetypescriptprefix [n:adventor] [TeXGyreAdventor] % gothic     URW Gothic L
\definetypescriptprefix [n:bonum]    [TeXGyreBonum]    % bookman
URW Bookman L
\definetypescriptprefix [n:chorus]   [TeXGyreChorus]   % chancery
URW Chancery L
\definetypescriptprefix [n:cursor]   [TeXGyreCursor]   % curier
Nimbus Mono L
\definetypescriptprefix [n:heros]    [TeXGyreHeros]    % helvetica
Nimbus Sans L
\definetypescriptprefix [n:pagella]  [TeXGyrePagella]  % palatino
URW Palladio L
\definetypescriptprefix [n:schola]   [TeXGyreSchola]   % schoolbook
Century Schoolbook L
\definetypescriptprefix [n:termes]   [TeXGyreTermes]   % times
Nimbus Roman No9 L


\definetypescriptprefix [f:adventor] [ag]
\definetypescriptprefix [f:bonum]    [bk]
\definetypescriptprefix [f:chorus]   [zc]
\definetypescriptprefix [f:cursor]   [cr]
\definetypescriptprefix [f:heros]    [hv]
\definetypescriptprefix [f:pagella]  [pl]
\definetypescriptprefix [f:schola]   [cs]
\definetypescriptprefix [f:termes]   [tm]


Then you need definitions for uc encoding:

\starttypescript [serif] [adventor,bonum,heros,pagella,schola,termes] [uc]
  \definefontsynonym [\typescriptprefix{n:\typescripttwo}-Regular]
   [file:texgyre\typescripttwo-regular]    [encoding=\typescriptthree]
  \definefontsynonym [\typescriptprefix{n:\typescripttwo}-Italic]
   [file:texgyre\typescripttwo-italic]     [encoding=\typescriptthree]
  \definefontsynonym [\typescriptprefix{n:\typescripttwo}-Bold]
   [file:texgyre\typescripttwo-bold]       [encoding=\typescriptthree]
  \definefontsynonym [\typescriptprefix{n:\typescripttwo}-BoldItalic]
   [file:texgyre\typescripttwo-bolditalic] [encoding=\typescriptthree]

  \definefontsynonym [\typescriptprefix{n:\typescripttwo}-Caps]
   [\typescriptprefix{n:\typescripttwo}-Regular]
  \definefontsynonym [\typescriptprefix{n:\typescripttwo}-ItalicCaps]
   [\typescriptprefix{n:\typescripttwo}-Italic]
  \definefontsynonym [\typescriptprefix{n:\typescripttwo}-BoldCaps]
   [\typescriptprefix{n:\typescripttwo}-Bold]
  \definefontsynonym
[\typescriptprefix{n:\typescripttwo}-BoldItalicCaps]
[\typescriptprefix{n:\typescripttwo}-BoldItalic]

  \definefontsynonym [\typescriptprefix{n:\typescripttwo}-Slanted]
   [\typescriptprefix{n:\typescripttwo}-Italic]
  \definefontsynonym [\typescriptprefix{n:\typescripttwo}-BoldSlanted]
   [\typescriptprefix{n:\typescripttwo}-BoldItalic]
\stoptypescript

% definitions for SmallCaps are wrong, since they're a feature in OT
fonts and that is a bit broken (or at least not stable enough) at the
moment.


And you also need the following:

% ec,texnansi,cs,el,qx,rm,t5,t2a,t2b,t2c,l7x
\starttypescript [serif] [times] [uc]

  \definefontsynonym [Times-Roman]       [TeXGyreTermes-Regular]
[encoding=\typescriptthree]
  \definefontsynonym [Times-Italic]      [TeXGyreTermes-Italic]
[encoding=\typescriptthree]
  \definefontsynonym [Times-Bold]        [TeXGyreTermes-Bold]
[encoding=\typescriptthree]
  \definefontsynonym [Times-BoldItalic]  [TeXGyreTermes-BoldItalic]
[encoding=\typescriptthree]
  \definefontsynonym [Times-Caps]        [TeXGyreTermes-Caps]
[encoding=\typescriptthree]

  \definefontsynonym [Times-Slanted]     [Times-Italic]
  \definefontsynonym [Times-BoldSlanted] [Times-BoldItalic]
\stoptypescript

Something strange is happening though. If you load these definitions
with XeTeX, it works OK. If one loads them with pdfTeX, it doesn't
work, since TeXGyreTermes-Regular isn't defined. When I think a bit, I
understand why it's not defined (because it's part of "termes" family,
not part of "times" family), but I don't understand why these
definitions work in XeTeX then.

Of course one could also say
  \definefontsynonym [Times-Roman] [file:texgyretermes-regular]    [encoding=uc]
but there should be a better way than to define the same font twice.

Is there any nice way to make "times" aware of "TeXGyreTermes-*" font names?

At the beginning I had problems with XeTeX as well since Times
resolved to 'Times Roman:mapping=tex-text', which resolved to the name
'mapping=tex-text' (a side effect of recent that should probably be
fixed somehow in the near future), but that problem disappeared later.
I had to temporary fix this (leftovers before TeXGyre family managed
to overload everything perhaps?):

\starttypescript[serif][times][uc]

\definefontsynonym[Times-Roman]     [Times Roman]       [encoding=uc]
\definefontsynonym[Times-Italic]    [Times Italic]      [encoding=uc]
\definefontsynonym[Times-Bold]      [Times Bold]        [encoding=uc]
\definefontsynonym[Times-BoldItalic][Times Bold Italic][encoding=uc]

\stoptypescript

Mojca

Re: xetext-pdftex differences

[Steffen Wolfrum]

Moica, thanks for your detailed support!

On Tue, 10 Apr 2007 21:27:47 +0200, Mojca Miklavec wrote:
>> Where do I go when I experience differences between context/pdftext 
>> and context/xetex?
> 
> What do you mean?

these:
 
>> For example with context/xetex I get ...
>> 
>> ! Undefined control sequence.
>> \stopframedtext ->\dostopframedtext
> 
...
>> ... and I do get "SS" instead of eszett ("ß") with the following:
> 
> Interesting effect ... (as if the encoding was unknown)


I have to admit that it I never understood which encoding to use when....
if, for example, I just write english/german/french then I could also use texnansi, this works perfect :

\usetypescriptfile[type-gyr]
\usetypescript[times][texnansi]
\mainlanguage[de]
\enableregime[utf]

\setupbodyfont[times,9pt]


> Which version of ConTeXt? It works OK here with 2007.03.22, but I need
> to try the new beta. In principle this should not happen, it's
> probably a bug or some other local problem.

it is version 2007.01.12 (didn't you recommend to stay with the january version?)


>> Whom to tell this?
> 
> Mailing list is the perfect place.


I was wondering whether the xetex or the context list is better as it is a mixed xetex/context-topic
and for Adam is in "paternity leave" I had no idea on which list someone continues.


> At the beginning I had problems with XeTeX as well since Times
> resolved to 'Times Roman:mapping=tex-text', which resolved to the name
> 'mapping=tex-text' (a side effect of recent that should probably be
> fixed somehow in the near future), but that problem disappeared later.
> I had to temporary fix this (leftovers before TeXGyre family managed
> to overload everything perhaps?):


right, "Times Roman:mapping=tex-text": This also happened to me when using ...

\usetypescriptfile[type-gyr]
\usetypescript[times][uc]

... that's why I used ...

\usetypescriptfile[type-gyr]
\usetypescript[times][ec]


... but probably I should have used "texnansi" (see above).


Steffen

Re: xetext-pdftex differences

[Mojca Miklavec]

On 4/11/07, Steffen Wolfrum <context@st.estfiles.de> wrote:
> Moica, thanks for your detailed support!
>
> On Tue, 10 Apr 2007 21:27:47 +0200, Mojca Miklavec wrote:
> >> Where do I go when I experience differences between context/pdftext
> >> and context/xetex?
> >
> > What do you mean?
>
> these:
>
> >> For example with context/xetex I get ...
> >>
> >> ! Undefined control sequence.
> >> \stopframedtext ->\dostopframedtext
> >
> ...
> >> ... and I do get "SS" instead of eszett ("ß") with the following:
> >
> > Interesting effect ... (as if the encoding was unknown)
>
>
> I have to admit that it I never understood which encoding to use when....

Me neither. Until I figured out that some of my letters were missing
in texnansi. texnansi & ec are 8-bit encodings used by "old" TeX
(pdfTeX) and only support 256 different characters. (you need to use a
different encodings for greek, cyrillic, vietnamese, ...)

The main problem is that XeTeX is not really backward compatible in
that respect: old fonts don't work since XeTeX thinks that the glyphs
are unicode-encoded, but they're in ec/texnansi/whatever other
encoding. If it nevertheless works, that is by a pure coincidence
(when you only need English letters ...).
In case that you would want to get the old fonts working with XeTeX,
you need to ask Jonathan. He said that it might be implemented, but
it's not really high on the priority list (it's just a problem during
transition state when 8'bit fonts still dominate in the TeX world).

LuaTeX will probably support those old encodings, but you would want
to use "uc" (unicode) encoding for both engines, otherwise it makes no
sense to use XeTeX instead of pdfTeX at all.

> if, for example, I just write english/german/french then I could also use texnansi, this works perfect :
>
> \usetypescriptfile[type-gyr]
> \usetypescript[times][texnansi]
> \mainlanguage[de]
> \enableregime[utf]
>
> \setupbodyfont[times,9pt]

That's because texnansi is based on "ansi" in the upper half of the
table and unicode is "by accident" the same in that area. So it's
really "coincedence" that it works. You should not use it that way
unless you have no better choice.

> > Which version of ConTeXt? It works OK here with 2007.03.22, but I need
> > to try the new beta. In principle this should not happen, it's
> > probably a bug or some other local problem.
>
> it is version 2007.01.12 (didn't you recommend to stay with the january version?)

It depends on what you want to do. The current version is indeed not
stable in that respect.

>
> >> Whom to tell this?
> >
> > Mailing list is the perfect place.
>
>
> I was wondering whether the xetex or the context list is better as it is a mixed xetex/context-topic
> and for Adam is in "paternity leave" I had no idea on which list someone continues.

All of your questions mentioned here belong to the ConTeXt mailing
list (XeTeX mailing list is mostly for XeLaTeX and low-level XeTeX
discussions) since it's really about the ConTeXt user-interface. The
only thing you could ask for on the XeTeX list is the question of
supporting ec/texnansi-encoded fonts (I prefer to wait for luaTeX than
to bother about those "soon-to-become-obsolete" additions).

>
> > At the beginning I had problems with XeTeX as well since Times
> > resolved to 'Times Roman:mapping=tex-text', which resolved to the name
> > 'mapping=tex-text' (a side effect of recent that should probably be
> > fixed somehow in the near future), but that problem disappeared later.
> > I had to temporary fix this (leftovers before TeXGyre family managed
> > to overload everything perhaps?):
>
>
> right, "Times Roman:mapping=tex-text": This also happened to me when using ...
>
> \usetypescriptfile[type-gyr]
> \usetypescript[times][uc]
>
> ... that's why I used ...
>
> \usetypescriptfile[type-gyr]
> \usetypescript[times][ec]
>
>
> ... but probably I should have used "texnansi" (see above).

You should try to get "uc" working if possible, but you need to add
some definitions from my previous mail. If you want to stick with
texnansi, you should better use pdfTeX.

Mojca

Re: xetext-pdftex differences

[Hans Hagen]

Mojca Miklavec wrote:
> LuaTeX will probably support those old encodings, but you would want
> to use "uc" (unicode) encoding for both engines, otherwise it makes no
> sense to use XeTeX instead of pdfTeX at all.
>   
luatex itself has still the pdftex machinery around, so you can use map 
files to handle things when you use for instance type one fonts.

ttf and otf are supported as wide fonts, but of course one needs to make 
sure that the unicode to glyph mapping is ok

in context mkiv i will intercept the encoding-prefixed fonts 
(texnansi-mojca-regular) so this is transparent for users by using some 
lua glue code but mainly because i read afm files instead of tfm files 
(so, texnansi-mojca will in fact use mojca.afm); non prefixed names will 
use the tfm file (this is pretty safe because the texnansi-* names were 
only there when proper afm files are also there; i mainly provide this 
features for existing typescripts)

in general, encodings are not something to worry about

not even uc as encoding is needed in mkiv
> All of your questions mentioned here belong to the ConTeXt mailing
> list (XeTeX mailing list is mostly for XeLaTeX and low-level XeTeX
> discussions) since it's really about the ConTeXt user-interface. The
> only thing you could ask for on the XeTeX list is the question of
> supporting ec/texnansi-encoded fonts (I prefer to wait for luaTeX than
> to bother about those "soon-to-become-obsolete" additions).
>   
indeed, there is no need to mess up xetex with obsolete encoding stuff; 
luatex will still have much pdftex code around so there the penalty is 
less, but eventually users have to move on

Hans

-----------------------------------------------------------------
                                          Hans Hagen | PRAGMA ADE
              Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
     tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com
                                             | www.pragma-pod.nl
-----------------------------------------------------------------

Re: xetext-pdftex differences

[Steffen Wolfrum]

On Wed, 11 Apr 2007 19:22:39 +0200, Hans Hagen wrote:

> luatex itself has still the pdftex machinery around, so you can use map 
> files to handle things when you use for instance type one fonts.


Does this mean for document with different fonts 
I will be able to use my good old mapfiles like

pads8r  AGaramond-Semibold       "TeXBase1Encoding ReEncodeFont" <8r.enc <pads8a.pfb
...

and also an opentype fonts like 

DecoTypeNaskh.ttf

together in one ConTeXt document, typeset with luatex?


Steffen 

Re: xetext-pdftex differences

[Steffen Wolfrum]

On Wed, 11 Apr 2007 17:44:18 +0200, Mojca Miklavec wrote:
> You should try to get "uc" working if possible, but you need to add
> some definitions from my previous mail. If you want to stick with
> texnansi, you should better use pdfTeX.

I have a document with German as the main language, defined ...

\usetypescriptfile[type-gyr]
\usetypescript[times][texnansi]
\mainlanguage[de]
\setupbodyfont[times,10pt] 

... and inserted Arabic quotes in Arabic AlBayan.ttf, defined  ...

\font\geeza="AlBayan" at 10pt


That's why it's typeset with XeTeX. But nevertheless texnansi for TeXGyreTermes seems to be fine (at least for now).


Thank you,

Steffen

Re: xetext-pdftex differences

[Hans Hagen]

Steffen Wolfrum wrote:
> On Wed, 11 Apr 2007 19:22:39 +0200, Hans Hagen wrote:
>
>   
>> luatex itself has still the pdftex machinery around, so you can use map 
>> files to handle things when you use for instance type one fonts.
>>     
>
>
> Does this mean for document with different fonts 
> I will be able to use my good old mapfiles like
>
> pads8r  AGaramond-Semibold       "TeXBase1Encoding ReEncodeFont" <8r.enc <pads8a.pfb
> ....
>
> and also an opentype fonts like 
>
> DecoTypeNaskh.ttf
>
> together in one ConTeXt document, typeset with luatex?
>   
indeed

- pads8r is a rather dump tfm file, so that goes through the normal tex machinery
- DecoTypeNaskh.ttf is an open type font and can be used directly 

\definefont[whatever][DecoTypeNaskh] 

should be enough


Hans 

-----------------------------------------------------------------
                                          Hans Hagen | PRAGMA ADE
              Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
     tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com
                                             | www.pragma-pod.nl
-----------------------------------------------------------------

Re: xetext-pdftex differences

[Hans Hagen]

Steffen Wolfrum wrote:
> On Wed, 11 Apr 2007 17:44:18 +0200, Mojca Miklavec wrote:
>   
>> You should try to get "uc" working if possible, but you need to add
>> some definitions from my previous mail. If you want to stick with
>> texnansi, you should better use pdfTeX.
>>     
>
> I have a document with German as the main language, defined ...
>
> \usetypescriptfile[type-gyr]
> \usetypescript[times][texnansi]
> \mainlanguage[de]
> \setupbodyfont[times,10pt] 
>
> .... and inserted Arabic quotes in Arabic AlBayan.ttf, defined  ...
>
> \font\geeza="AlBayan" at 10pt
>
>
> That's why it's typeset with XeTeX. But nevertheless texnansi for TeXGyreTermes seems to be fine (at least for now).
>   
technically one can define a virtual font which uses TeXGyreTermes as base, but has additional glyphs taken from AlBayan; I do have experimental code for that but have to cook up an interface (not that hard, i only need time and motivation); that way you dont't need to switch to geeza in between 

\ctxlua {
    fonts.define.methods.install("weird", {
{ "include-method", "base" } ,
{ "copy-char"     , "lmroman10-regular"     ,     65,     66 } ,              { "copy-range"    , "lmsans10-regular"      , 0x0100, 0x01FF } , 
{ "copy-range"    , "lmtypewriter10-regular", 0x0200, 0xFF00 } ,
{ "fallback-range", "lmtypewriter10-regular", 0x0000, 0x0200 }   
    })
}

\starttext

\startTEXpage

\font\test={myfont@weird}
\test \eacute \rcaron \adoublegrave \char"0200
AB\char65\char65

\stopTEXpage

Along these lines ... but then properly interfaced -) 


Hans 

-----------------------------------------------------------------
                                          Hans Hagen | PRAGMA ADE
              Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
     tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com
                                             | www.pragma-pod.nl
-----------------------------------------------------------------

Re: xetext-pdftex differences

[Steffen Wolfrum]

On Thu, 12 Apr 2007 12:46:51 +0200, Hans Hagen wrote:
> Steffen Wolfrum wrote:
>> On Wed, 11 Apr 2007 17:44:18 +0200, Mojca Miklavec wrote:
>>   
>>> You should try to get "uc" working if possible, but you need to add
>>> some definitions from my previous mail. If you want to stick with
>>> texnansi, you should better use pdfTeX.
>>>     
>> 
>> I have a document with German as the main language, defined ...
>> 
>> \usetypescriptfile[type-gyr]
>> \usetypescript[times][texnansi]
>> \mainlanguage[de]
>> \setupbodyfont[times,10pt] 
>> 
>> .... and inserted Arabic quotes in Arabic AlBayan.ttf, defined  ...
>> 
>> \font\geeza="AlBayan" at 10pt
>> 
>> 
>> That's why it's typeset with XeTeX. But nevertheless texnansi for 
>> TeXGyreTermes seems to be fine (at least for now).
>>   
> technically one can define a virtual font which uses TeXGyreTermes as 
> base, but has additional glyphs taken from AlBayan; I do have 
> experimental code for that but have to cook up an interface (not that 
> hard, i only need time and motivation) ...


No no, stop!!! Don't do it the hard way. 

ConTeXt has already such handy ways to fulfill my needs.
Switching between the two fonts is perfectly fine for me. 

Thank you, Steffen.