ExTeX
  Development     Documentation     Download    


Objectives

Modularity

εχTEX's object-oriented design (see figure) includes some main modules that act without knowledge about the functioning of the other modules (encapsulation). Thus the frontend with tokenisation and the expansion of tokens, up to the creation of nodes are exchangeable, independently from the typesetter. The same applies to the backend that generates the output. A special role plays the font module that provides font information to the typesetter as well as to the output module.

Inside the main modules other subdivisons are established, making subtitution or enhancement of single aspects of a function possible. So the font module may be extended to new formats by adding or replacing a font reader, without directly affecting other parts of the program. Likewise new algorithms may be integrated into the typesetter that is responsible for line breaking and the structure of the page.

Compatibility

εχTEX will provide all of TEX's primitives. Since syntax and function of some primitives (e.g. \input) leads to undesirable constraints on modern systems, supplemental primitives will be designed and implemented in these cases. The old variants will clearly be marked obsolete. In some cases this has been done already. Hardly understandable peculiarities of TEX can be abandoned in favour of a clean implementation. Moreover the behaviour in case of an error will not be emulated completely equally. Eventually εχTEX shall be able to typeset existing documents visually identically to TEX.

εχTEX will not be compatible regarding the strict separation of iniTEX and virTEX. So far, loading of hyphenation patterns and creation of format files is possible only for iniTEX. εχTEX on the other hand shall be able to load hyphenation patterns globally at any time, i.e. it shall not be limited to hyphenation tables that are stored in the format files. Therefore it will be possible to load only those hyphenation tables at runtime that are actually needed. εχTEX also has the ablility to create a format under particular conditions, namely outside of any group.

From ε-TEX mainly those facilities will be adopted whose main effect is not limited to the logging messages. The additional tracing primitives are not adopted to the vocabulary of εχTEX because εχTEX will provide its own progressive facilities for debugging purposes. The facilities of ε-TEX to toggle the direction of writing will be omitted, too. These can be realised with the help of Ω-compatible primitives on macro level. Thus compatibility to real existing ε-TEX documents should be obtained.

From Ω the extensions regarding the fonts will be taken, as well as the feature of setting the direction of writing. However εχTEX will not adopt the encoding functions of Ω, but instead realise 16-bit capability with its own easily comprehensible methodes. Here established basic facilities of Java will be used.

There are two kinds of pdfTEX features: those controlling PDF output, and those providing additional typographic features. The latter shall be widely adopted. Here the problem arises that in pdfTEX these primitives are prefixed with »pdf«, too. Hence new names and partially other primitives will replace them in εχTEX. As above, macros may be used as the connective link. In most cases, e.g. with LATEX packages like graphics or hyperref, a change to a new driver is recommended. This all the more, as the way to implement capabilities that are strongly depenent on the backend is different inεχTEX.

New Features

Although εχTEX is designed so that new facilities can easily be added to every main module – actually on the level of primitives – εχTEX will right from start be equipped with some innovations. One of these is the handling of different encodings of input files. At this εχTEX uses Java's built-in facilities to transfer read characters into an internal 32-bit coding.

As several of TEX's primitives are limited to 8-bit, and Ω to 16-bit, new appropriate equivalents will be defined, e.g. for \mathcode. On the other hand, \char will be able to handle values greater than 255 directly. Actually this is an incompatibility to TEX, but since TEX would report an error in this case, this behaviour is compatible to existing error-free documents.

Expansions that do not depend on a specific output format, but are – even in various ways – supported by different output formats, will have their own primitives. Examples for this are the begin and end of links as well as their targets./

TEX knows elements that do not deal with typography of the page, but they must be processed when it is written out (\shipout). Examples for this whatsits are \openout, \closeout and \write, always without being prefixed with \immediate. Especially these should not have any effect on page breaking, which can't be guaranteed by TEX as is. This deficency shall be improved in εχTEX, despite arising incompatibilities.

Among the whatsits, the \special primitive plays a special role. Depending on its meaning, it may have effects on line or page breaking (e.g. if a graphic is inserted), or not (e.g. if colour is changed). Apart from the fact that εχTEX will support colours directly, εχTEX will have a new primitive for extensions that depend on the output format. When encountering this primitive, the typesetter will demand the decision whether the element affects line or page breaking from the backend (see IObject and Boxsize in the figure). For example, a PDF backend may return the size of an image that was inserted by this primitive.

Medium-term εχTEX shall provide graphical elements like lines of arbitrary gradient, ellipses, bezier curves etc. Also graphical transformations like rotation, reflexion, or scaling shall be implemented.

Recent Publications
Minimal? – Vom Testen eines Textsatzsystems, Die TeXnische Komödie 3/2006.
Namespaces for εχTEX, EuroTeX 2005 (Slides)
εχTEX – Under the Hood, EuroTeX 2005 (Slides)
Contact Address
Sponsored by
DANTE e.V.
Hosted by
BerliOS
© 2014 The εχTEX Group