Following is a set of html directives and restrictions used for the publishing of "collections" of "chapters", "papers" or "articles", when such collections are to be distributed as both "web-pages", and as a stand-alone computer-media resident publications (on floppy disks, CD's, etc.). Such "publications" are read using "web-browser" programs (for example Mosaic, NetscapeNavigator, InternetExplorer, Opera, etc.) This text assumes that the reader is familiar with the essentials of the operating system independent computer file systems and with the basics of web publishing (html, bit-mapped graphics).

One such "collection" can be thought of as the equivalent of one "bound volume" in the paper world (book, technical manual, magazine, conference proceedings...). In further text it will be referred to as a "book". As already stated, a book is assumed to consist of "articles". It is further assumed that multiple books will reside and be distributed on a single computer disk. This document assumes that there will be one "decision-maker" (person or institution) that acts as a "book publisher" and a (possibly large) number of others that are considered "article authors".

This document addresses two principal issues: how to make more than one such book co-resident on a single disk, and how to internally organize a book, so that the manipulation of the material on a computer is as simple as possible, and can be "read" using as wide a range of computers and browsers as practical.

One book is completely contained in a single, "flat" directory. ("Flat" means there are no subdirectories). The name of the book is a directory name, and it consists of lower-case letters, numbers and the underscore ("_") character, with the total count of less than 64. The starting character must be a letter. This standard does not control the book name space - which means it provides no rules that would guarantee such names would be unique. However, if the names are as complete and as descriptive as the names of the paper publications commonly are, and if they end with the publication date in 'yyyymmdd' form, the likelihood of name collision will be greatly reduced.

A book consists of a "table of contents", and "articles". Since all book content is assumed to be read in a browser, the table of contents will be an index.html file. The book publisher will assign a unique string, starting with a lower case letter, of up to 16 characters, conforming to a file-base-name convention, to each article. This string will be referred as "article name". (Note, however, that the reader will not normally see this name). Each file that "belongs" to an article has a file base name which starts with the "article name". Similar to the directory names, file name are strings of less than 64 characters. Unlike directory (or book) names, file names will include exactly one period (".") character; separating the file base name from the extension. Each article will have one file whose name consists of the article name and an ".html" extension: this is the html file that will be linked to the html "hypertext reference" from the article title in the book table of content (index.html file).

This completes the set of rules that this "standard" specifies. However, the book publisher will usually add additional rules which the article authors are expected to follow. The purpose of such rules would be to ensure that, for instance, all articles can be read on a given level of browsers, that there is some guaranteed level of service in either web or media distribution, that all articles follow the same typography and presentation style, and so on. What follows is an example of such rules for article author(s):

1. All file names will be unique, starting with the article name. Files must be of one of the following types (defined by their extensions): .html, .gif, .png and .jpeg . There will be one file named artclname.html (where artclname stands for the article name assigned by the book publisher); this file will be the main text content of the article in the form of an html document. It will be the only file with META tags, and it will have a <title> tag which gives a full title of the article. The article title will also appear as an <h2> header, all other headers within the document (if any) will be of lower (<h3>, <h4> etc.) levels.

2. An article can have at most 8 .html files and at most 16 graphical files, totaling no more than 50 KB for all .html files combined, and no more than 200 KB of combined graphical (.gif, .png, and .jpeg) files.

3. Html constructs used must conform to the 3.2 level. Standard html (UNIX-like) line endings are preferred. There are further restrictions, as noted below.

4. Individual image size must not exceed 400 pixels in any dimension (width or height). .gif or (preferably) .png file format will be used for vector (computer generated) graphics, .jpeg for "dense-field" (photographs, etc.) graphics. .gif and .png files must use "web-safe" ("WEBCOLORS") palette. All images must include a meaningful description in an ALT tag.

5. Hypertext references which are internal to the article can be "implicit" in the text and must use relative URL's; external references must be clearly described as net URL's in the text, and must be absolute URL's. Internal references may not open a separate browser window; external ones must.

6. Following html features must not be used: background color control; background graphics; sound; link appearance modification; browser window size or margin control; text-in-graphics; frames; server-side processing of any kind (CGI etc.); forms; plug-ins; java, java-script VB-script or style sheets; <font>, <font-size> or <blink> tags and text color control.

7. Following html features are tolerated but discouraged: more than one .html file per article; animated .giff's; mail links; client-side imagemaps and paragraph or header indentation or justification control.

Finally, it is suggested to the article authors to test their documents with browsers that put a great deal of emphasis on enforcing the html standard and that restrict the functionality in order to improve loading speed and responsivness. Under Win32 operating systems this can be done with Opera (http://www.operasoftware.com/); under Unices with GNU Help Browser or similar. Such browsers will often demonstrate html errors even in instances that are quietly (but sometimes inconsistently) tolerated by mainstream browsers.

If any of the web-publishing concepts mentioned in this document need further explanation, the suggested reference is: Web Design in a Nutshell - A Desktop Quick Reference, J. Niederst, O'Reilly, 1999.