<\/p>\n
In a typical document, there is a hierarchy of sections. Each section contains a heading, which is displayed more prominently than the section text. <\/p>\n
In this article, I describe three separate issues with the handling of sections and headings in modern typesetting tools. I propose a dull but important change to markup languages to deal with these issues. This proposal is not particularly innovative, but just better engineered.<\/p>\n
Then I mention a new problem that I only just noticed, and propose an innovative new feature that would be way cool.<\/p>\n
During the editing of a document, the section hierarchy can be quite fluid. Sections can be moved around the document, or promoted or demoted in place.<\/p>\n
Microsoft Word makes moving these sections relatively easy (especially if you configure a custom toolbar to always display the \u00e2\u20ac\u0153Promote\u00e2\u20ac\u009d, \u00e2\u20ac\u0153Demote\u00e2\u20ac\u009d and \u00e2\u20ac\u0153Demote To Body Text\u00e2\u20ac\u009d icons \u00e2\u20ac\u201c they normally only appear in Outline mode, but are universally useful.)<\/p>\n
However, the common markup languages do not make this easy at all: HTML, LaTeX and MarkDown all require you to visit every heading in the text being moved, and correct the heading level. <\/p>\n
For example, if I decide to demote a large section of text in HTML, I would need to change all the <H1> tags to <H2>, but first change all the <H2> tags to <H3> etc. In LaTeX, I would need to change all the \\section{} tags to \\subsection{}, etc.<\/p>\n
Compare that to bulleted lists, where it is possible to move list items around without renumbering everything. The browser can work out the appropriate bullet level from context.<\/p>\n
I once wasted several days\u00e2\u20ac\u2122 development time in producing code that marked up a stack-dump in LaTeX tables. It worked fine in the unit tests, but the first time that I tried a \u00e2\u20ac\u0153substantial\u00e2\u20ac\u009d program (for small values of substantial: it calculated the oddness of 2<\/a>), LaTeX complained that I had blown its limit of sub-sub-sub-tables, and suggested that I modify its source code to increase the arbitrary limit and recompile it. I declined, and re-wrote the stack dump into raw PostScript. Having arbitrary limits in your software should be avoided; it just means you annoy people at arbitrary times. <\/p>\n I had a problem with tables, but the same applies to sections too: In HTML and LaTeX, there are arbitrary limits to the depth of the section hierarchy. HTML has <H1> to <H6>. LaTeX runs out of puff after subsubsections<\/a>. Both of these are perfectly sufficient for 99% of documents, but why introduce any limit at all?<\/p>\n One method to emphasise the hierarchical nature of a document is to modify the style of the text that appears in a section. Here are some examples: deeply nested sections might be indented further to the right; the first word of a chapter might use drop-caps.<\/p>\n This is especially important if you want to be able to be able to add more text to the parent section beneath the text of the sub-section.<\/p>\n None of Word, HTML, LaTeX and MarkDown offers the ability to do this easily.<\/p>\n I would propose a markup like so:<\/p>\n This is a better model than the existing systems, because it fixes all the above issues:<\/p>\nIssue 3: Body Text\/Heading Associations<\/h3>\n
A Dull But Important Proposal<\/h3>\n
<SUBSECTION TITLE=\"Vertebrates\">\n <SUBSECTION TITLE=\"Mammals\">\n Mammals are warm-blooded and suckle their young. \n They are often cute and cuddly.\n <\/SUBSECTION>\n <SUBSECTION TITLE=\"Reptiles\">\n Reptiles are slimy and icky. Ewww! Don\u00e2\u20ac\u2122t cuddle them.\n <\/SUBSECTION>\n There are other types of vertebrates, too, \n but I don\u00e2\u20ac\u2122t want to mention them here.\n <\/SUBSECTION>\n<\/code><\/pre>\n