LaTeX vs. MiKTeX: The levels of TeX

A friend once asked us, “Should I use LaTeX or MiKTeX?” In various guises, this is a common question, seemingly innocent, but actually betraying a fundamental confusion about the levels of operation in the TeX world. As a further confusion, the word “TeX” can be used to refer to any of a myriad of items at any level. Starting at the top:

1. Distributions: MiKTeX, TeX Live, W32TeX, … These are the large, coherent collections of TeX-related software to be downloaded and installed. When someone says “I need to install TeX on my machine”, they're usually looking for a distribution.
2. Front ends and editors: Emacs, vim, TeXworks, TeXShop, TeXnicCenter, WinEdt, … These editors are what you use to create a document file. Some (e.g., TeXShop) are devoted specifically to TeX, others (e.g., Emacs) can be used to edit any sort of file. TeX documents are independent of any particular editor; the TeX typesetting program itself does not include an editor.
3. Engines: TeX, pdfTeX, XeTeX, LuaTeX, … These are the executable binaries which implement different TeX variants. In short:
   • pdfTeX implements direct PDF output, along with a variety of programming and other extensions.
   • XeTeX does the above, and also supports Unicode natively, OpenType and TrueType fonts, access to system fonts, …
   • LuaTeX does all the above, and provides access to many internals via the embedded Lua language. Thus it is by far the most programmable engine.
   • [e][u]pTeX provide full support for Japanese typesetting.
   There are other engines, but the above are by far the most commonly used nowadays.
4. Formats: LaTeX, plain TeX, OpTeX, … These are the TeX-based languages in which one actually writes documents. When someone says “TeX is giving me a mysterious error”, they usually mean a format.
5. Packages: geometry, lm, … These are add-ons to the basic TeX system, developed independently, providing additional typesetting features, fonts, documentation, etc. A package might or might not work with any given format and/or engine; for example, many are designed specifically for LaTeX, but there are plenty of others, too. The CTAN sites provide access to the vast majority of packages in the TeX world; CTAN is generally the source used by the distributions.

Output formats

TeX source files can be typeset into several different output formats, depending on the engine. Notably, the pdfTeX engine (despite its name) can output both DVI and PDF files.

At a high level, the output format that gets used depends on the program you invoke. If you run latex (which implements the LaTeX format), you will get DVI; if you run pdflatex (which also implements the LaTeX format), you will get PDF.

No TeX engine implements native HTML output, but it is still possible to get HTML, XML, etc., output:

• The tex4ht program can be run (e.g., htlatex, make4ht). TeX4ht uses TeX behind the scenes, so user macros, etc., are generally recognized. There are many possible output formats, including Office XML. TUGboat article on tex4ht, and more recent article.
• The lwarp LaTeX package causes LaTeX to output HTML5. It hooks into many packages to create the html. TUGboat article on lwarp.
• The LaTeXML Perl program independently parses LaTeX documents and generates many output formats.

ConTeXt

ConTeXt is a special case, straddling levels. It contains a format at the level of plain TeX and LaTeX, but unlike the other formats, it is invoked via a separate program (e.g., context) which then runs a TeX engine. This makes it possible to support a wide array of advanced features, such as integrated graphics and XML input, since the control program can determine the flow of processing.

References

Of course, this short web page is only a brief introduction to the basics. Here are some pointers to further information.

• Getting started with TeX, the TUG page giving an introduction to various parts of the TeX world.
• Components of TeX, Joachim Schrod's technical article describing many of the relationships hinted at here in more detail. It was written many years ago, but the relationships still hold.
• A First Set of LaTeX Resources, Jim Hefferon's document recommending LaTeX packages for many common tasks.
• Things with “TeX” in the name, FAQ entry.
• The TeX Family in 2009, an article published in AMS Notices by Jim Hefferon and Karl Berry, giving an overview of the TeX engines, graphics, hypertext, presentations, fonts, etc. Although now more than a decade old, it still reasonably describes the current state of things.