Prev: Intel 80386 Programmer's Reference Manual
Next: Preface

About This Manual


The original source of this manual (INTEL386.TXT), as far as I can tell, is an official Intel source document. I found the original document somewhere on the net (I don't remember where). You can get the original source document from here: (212K)
(link current as of 3/97).

This hypertext version of the source document, obviously, is not an official Intel document or product. You can download the entire hypertext-ed manual from here: (2??K)
(link current as of 3/97).


I converted the manual to html because:
(1) I needed a basic 80x86 instruction reference,
(2) the latest and greatest format of the manuals are huge PDF files (on the order of 3-4meg each) and are a pain to sift thru with Acrobat,
(3) I thought it would be a fun project.

Character Set

Because the original document used the extended PC character set, I decided to stick with it as much as possible. At one point, I converted everything to standard ASCII, but found that some characters were muddled in the shuffle. So ... to properly display the tables and some of the other characters, your browser must be using the MS LineDraw character set. In Netscape 2.0, use Options/Preferences/Fonts/Fixed Font-Choose Font. Select MS LineDraw. I use a font size of 10.

If your browser is set correctly, the figures below will be (1) A box with all double lines, (2) a box with horizontal double lines and vertical single lines, (3) a box with horizontal single lines and vertical double lines, (4) a box with all single lines. If not, then you should still see the shapes, but the lines will be drawn with the wrong characters.
  (1)      (2)      (3)      (4)
+--+--+  +--+--+  +--+--+  +--+--+
|  |  |  |  |  |  |  |  |  |  |  |
|--+--|  |--+--|  |--+--|  |--+--|
|  |  |  |  |  |  |  |  |  |  |  |
+--+--+  +--+--+  +--+--+  +--+--+
An unfortunate side effect of using the line drawing character set is that the bullets in lists appear as a large line drawing character. One character that couldn't be translated to my statisfaction was the solid down arrow, ASCII #31. I ended up using the exclamation mark for it. This character appears in some of the figures in the document.

Conversion Process

The conversion process was rather simple in concept, but ended up taking a lot longer than I expected (mainly because I had to write all the software to do it).

With a cursory inspection of the manual, I found it to be divided into 6 parts; everything before Part I, Parts I, II, III, IV, and the appendices. Within each part are either specific sections, or Chapters. The chapters are divided into sections, and the instruction set was in its own world. Also, I noticed the use of the extended character set (see above); the line drawing characters and even some characters between #1 and #31 ASCII.

The first task was seeing how the extended characters would display in Netscape. Not real good. I poked around a bit, and found out that I could set the character set to MS LineDraw, which fixed most of the problem, but the low end extended characters still wouldn't display correctly. I ended up developing the TXLATE program to convert the extended characters to a pseudo-ASCII equivalent.

The next task was to split the file into pieces. Since the original file is 878K, the goal was to have each piece a manageable size, say around 20K. By breaking each chapter down into sections, this was achieved except for about a dozen cases (less than 10% of the files). For this job, the TSPLIT program was developed. This program reads a command file, and splits up input files into output files based on strings. Simple strings didn't cut it, so I had to convert the regexp(3) routines from C into Pascal. This also caused me to develop the GREP program, and a program called TGREP which is a subset of GREP.

Chapter 17, the Instruction Set, provided a unique problem. Section 17.2 actually contained all the instructions, making this a huge section. I elected to add a new section, 17.3, which only contained instructions, and to group the instructions alphabetically into sub-sections, i.e. 17.3.A contains the 'A' instructions.

Once the original file was in pieces, each piece had to be converted into HTML with tags and links added to it. The first task was to index the files. The TINDEX program was developed for this. This program reads an input file, indexes it based on a command file containing search and pattern strings, and writes an output file consisting of the file name, tag, and source line. The program also assumes that the first line of each file is the title.

Once all the files are indexed, they needed to be converted from TXT to HTM with appropriate markup codes added. The THTML program was developed for this purpose. THTML reads an input file, adds appropriate HTML information, and checks for references to the input file in an index file. Matching lines are marked with A NAME= tags.
After I had an initial run, I noticed there was some things lacking. I still needed to create the table of contents, figures, and tables pages; and I also wanted to have a chapter index at the start of each chapter. The THREF program was developed for this. This program reads an index file (output by TINDEX) and generates A HREF= entries. By TGREPing the index file and piping the output to THREF, simple indexes could now be created. The TREPLACE, TINDENT, and TBLANK programs were developed to polish the results.

Do It Yourself?

If you'd like to try this yourself, you can download a ZIP file containing the batch file (I386.BAT) and other related files from here: (9K)
(link current as of 3/97).

You'll also need the following utility programs:
Note that these programs are not specific to this conversion, but are a general set of utilities for the conversion of somewhat structured text files.

They are available from here:
(link current as of 3/97).

Comments, etc. to:
Lance Costanzo
Prev: Intel 80386 Programmer's Reference Manual
Next: Preface