Zdzislaw Meglicki <[email protected]> writes: >> What do you mean by DTD XML? All the standard XML data files follow a >> DTD (document type definition). > > This is what Tralics papers say and then add "similar to TEI XML". Perhaps > there are some more relaxed XMLs around that are not DTD. Or perhaps > theirs is a unique XML that is similar to TEI and also DTD compliant. > Something like this, I guess.
I cannot parse the phrase "dtd xml". I expect that both tralics and LaTeXML have their own xml document types. An XML document type may be defined by a DTD or by what is called a schema. Essentially it is a markup vocabulary not unlike LaTeX with rules for what can be used where. Also it seems that Connexions has its own xml document type. For this see http://cnx.org/content/m9000/latest/ My guess is that this document type provides the source language for what appears online at Connexions. > Here's my salesman pitch: on-line education is coming. It's become a > hot issue in the US already with top universities actively > engaged. So we need TeX to produce XML (or HTML5 at the very least) > and it needs to be done by TeX itself, that is, from within the > official TeX distribution suite. Yes, this is an important distinction. It is too restrictive to insist that a TeX engine do all such processing. But given that one wants LaTeX source -- or at least LaTeX-like source -- it would be good if there was TeX community maintenance and sponsorship. (Again see my TUG 2010 talk about LaTeX profiles.) > The approach taken by Tralics and LaTeXML is to work from the LaTeX source > and convert "the meaning" of constructs encountered to XML/MathML. > But for this to work, there must be a special substitute for every LaTeX > package included in the source. The resulting conversion may break > on silly things like, say, direct TeX commands, for example LaTeXML > does not know about "\mathinner", so I can't include the braket package. > Tralics alters TeX semantics in subtle ways, e.g., with equation numbering > and references, and with spacing around brackets. Exactly. For a given suite of documents one needs a fixed list of commands, but it would be too much to ask for an all-encompassing universal list of commands. > To convert from dvi instead can give the system more flexibility, and > we already have dvipdfm and similar example applications that show > how to read the file and how to construct the output in, say, page > description language such as PDF. And this is exactly what TeX4ht > is trying to do, so it looks like a good way forward. Not really. Tex4ht DVI files are very non-standard. Standard DVI admits translation only to printer or printer-like languages. -- Bill
