xml2rfc v1.35

Table of Contents

1.  Introduction
2.  Requirements
3.  Testing
3.1.  Testing under a windowing system
3.2.  Testing without a windowing system
4.  Next steps
4.1.  Processing Instructions
4.1.1.  Option Settings
4.1.2.  Include Files
5.  The Page Model
6.  Additions to RFC 2629
6.1.  Extra Attributes
6.2.  Typed-Artwork Interpretation
7.  Limitations of xml2rfc
8.  References
A.  MacOS 9 Installation (courtesy of Ned Freed)
B.  rfc2629.xslt (courtesy of Julian Reschke)
C.  MS-Windows XP/Cygwin Installation (courtesy of Joe Touch)
D.  A Special Thanks
E.  Copyrights
§  Authors' Addresses

1.  Introduction

This is a package to convert memos written in XML to the RFC format.

If you don't want to install any software, you can use the web-based service.

2.  Requirements

You need to have Tcl/Tk version 8 running on your system. Tcl is a scripting language, Tk is Tcl with support for your windowing system.

To get a source or binary distribution for your system, go to the Tcl Developer Xchange website and install it. If you get the binary distribution, this is pretty simple.

Of course, you may already have Tcl version 8. To find out, try typing this command from the shell (or the “MS-DOS Prompt”):

    % tclsh

If the program launches, you're good to go with Tcl version 8.

If you are running under a windowing system (e.g., X or MS-Windows), you can also try:

    % wish

If a new window comes up along with a “Console” window, then you're good to go with Tk version 8.

Finally, you may notice a file called xml2sgml.tcl in the distribution. It contains some extra functionality for a few special users — so, if you don't know what it is, don't worry about it...

3.  Testing

Now test your installation.

3.1.  Testing under a windowing system

Type this command from the shell:

    % xml2rfc.tcl

A new window should come up that looks like this:

Fill-in the blanks and click on [Convert].

3.2.  Testing without a windowing system

Type this command from the shell:

    % tclsh

If the program launches, type this command to it:

    % source xml2rfc.tcl

and you should see these five lines:

    invoke as "xml2rfc   input-file output-file"
           or "xml2txt   input-file"
           or "xml2html  input-file"
           or "xml2nroff input-file"
           or "xml2unpg  input-file"

4.  Next steps

Read the 2629bis document. In particular, Section 3 has some good information.

4.1.  Processing Instructions

A processing instruction contains directives to an XML application. If you want to give directives to xml2rfc, the processing instructions (PIs) look like this:

    <?rfc keyword='value'?>

Of course, if you like the default behavior, you don't need any behavior-modifying directives in your input file! Although xml2rfc supports putting several attribute-like directives in one PI, be warned that there are issues in doing this for a non-include-file directive following an include-file directive (Include Files). It is good practice to always surround the value with either single or double quotes.

4.1.1.  Option Settings

The list of valid keywords are:

Remember that, as with everything else in XML, keywords and values are case-sensitive.

With the exception of the needLines, typeout, and include directives, you normally put all of these processing instructions at the beginning of the document (right after the XML declaration).

4.1.2.  Include Files

xml2rfc has an include-file facility, e.g.,

    <?rfc include='file'?>

xml2rfc will consult the XML_LIBRARY environment variable for a search path of where to look for files. (If this environment variable isn't set, the directory containing the file that contains the include-file directive is used.) The file's contents are inserted right after the PI. Putting non-include-file directives (especially needLines ones) after an include-file one in the same PI may not work as expected because of this. Remember that file names are generally case-sensitive and that an input file that is distributed to the outside world may be processed on a different operating system than that used by its author.

You can also have xml2rfc set the XML_LIBRARY environment variable directly, by creating a file called .xml2rfc.rc in the directory where your main file is, e.g.,

global env tcl_platform

if {![string compare $tcl_platform(platform) windows]} {
    set sep ";"
} else {
    set sep ":"
}

if {[catch { set env(XML_LIBRARY) } library]} {
    set library ""
    foreach bibxmlD [lsort -dictionary \
                           [glob -nocomplain $HOME/rfcs/bibxml/*]] {
        append library $sep$bibxmlD
    }
}

set nativeD [file nativename $inputD]
if {[lsearch [split $library $sep] $nativeD] < 0} {
    set library "$nativeD$sep$library"
}

set env(XML_LIBRARY) $library

There are links to various bibliographic databases (RFCs, I-Ds, and so on) on the xml2rfc homepage.

5.  The Page Model

The html rendering engine does not need to have a tightly defined page model.

The txt and nr rendering engines assume the following page model.

Each line has at most 72 columns from the left edge, including any left margin, but excluding the line terminator. Every output character is from the ASCII repertoire and the only control character used is the line-feed (LF); the character-tabulation (HT) character is never used.

Each page has the following lines (in 1-based numbering, as reported to the user, but contrary to xml2rfc's internal 0-based numbering):

 1: header line (blank line on first page)

 2: blank line

 3: blank line

 4: 1st line of content

...

51: 48th line of content

52: blank line

53: blank line

54: blank line

55: footer line

56: form-feed character (followed by line terminator)

Once processed through nroff and the fix.sh script (from 2-nroff.template), the nr output differs from this in two ways. It has three extra blank lines (that could be numbered -2, -1, and 0, for a total of six) at the very beginning of the document (so the first page is that much longer). It also has no line terminator following the very last form-feed character of the file. These differences originate in the design of the fix.sh script.

Header and footer lines each have three parts: left, center, and right.

6.  Additions to RFC 2629

A few additions have been made to the format originally defined in RFC 2629. In particular, Appendix C of the 2629bis document enumerates the additions.

6.1.  Extra Attributes

In addition, xml2rfc recognizes the undocumented src, alt, width, and height attributes in the figure and artwork elements, but only if HTML is being generated. Here are two examples, one for each case.

Here, the attributes are added to the artwork element.

<figure>
    <preamble>This is the preamble.</preamble>
    <artwork src='layers.png'
             alt='[picture of layers only]'>
.-----------.
| ASCII art |
`-----------'
</artwork>
    <postamble>This is the postamble.</postamble>
</figure>

In this case, the preamble and postamble elements are kept and an img tag is placed in the HTML output to replace the whole artwork element and its textual drawing, which are ignored.

Here, the attributes are added to the figure element.

<figure src='layers.png'
        alt='[picture of layers and explanation]'>
    <preamble>This is the preamble.</preamble>
    <artwork>
.-----------.
| ASCII art |
`-----------'
</artwork>
    <postamble>This is the postamble.</postamble>
</figure>

In this case, an img tag is placed in the HTML output to replace the whole contents of the figure element (the preamble, artwork, and postamble inner elements and the textual drawing itself) which are ignored.

xml2rfc also recognizes an undocumented align attribute (with possible values left, center, or right) in the figure and artwork elements. It affects the whole content of the targeted element for all types of generated output. Its default is inherited from the parent of its element.

6.2.  Typed-Artwork Interpretation

The artwork element from RFC 2629 supports an optional type attribute. While most possible values are just ignored, including the special case where the attribute is unspecified or just empty, some values are recognized. In particular, type='abnf' can be used if the artwork contains an Augmented Backus-Naur Form (ABNF) syntax specification [3] (Crocker, D. and P. Overell, “Augmented BNF for Syntax Specifications: ABNF,” October 2005.). As a special extension in its behavior, xml2rfc will attempt to validate the syntax and colorize the HTML output of ABNF, since it is so widely used in RFCs. It does this colorizing by relying on the full parsing it does right before, not on a quick and partial (e.g., line-by-line) pattern-based hack. ABNF is the only artwork type to benefit from this kind of internal support at this time. If the strict rfc-PI directive is activated, invalid ABNF content will cause xml2rfc to abort with an error message. Omitting the type attribute altogether is the obvious way to avoid having this validation and colorizing performed.

For example (to be viewed in HTML):

char-val       =  DQUOTE *(%x20-21 / %x23-7E) DQUOTE
                       ; quoted string of SP and VCHAR
                          without DQUOTE

num-val        =  "%" (bin-val / dec-val / hex-val)

bin-val        =  "b" 1*BIT
                  [ 1*("." 1*BIT) / ("-" 1*BIT) ]
                       ; series of concatenated bit values
                       ; or single ONEOF range

dec-val        =  "d" 1*DIGIT
                  [ 1*("." 1*DIGIT) / ("-" 1*DIGIT) ]

hex-val        =  "x" 1*HEXDIG
                  [ 1*("." 1*HEXDIG) / ("-" 1*HEXDIG) ]

prose-val      =  "<" *(%x20-3D / %x3F-7E) ">"
                       ; bracketed string of SP and VCHAR
                          without angles
                       ; prose description, to be used as
                          last resort

This is from the original RFC on ABNF [2] (Crocker, D., Ed. and P. Overell, “Augmented BNF for Syntax Specifications: ABNF,” November 1997.), with its minor mistakes in manually folded comment lines purposely left intact, for illustration. Since the result is still valid ABNF (but incorrect with respect to what was intended), this showcases how colorizing might give a human author (or editor or reader) a better chance to spot the three mistakes (and correct them, e.g., with extra semicolons, as has been done in a more recent version [3] (Crocker, D. and P. Overell, “Augmented BNF for Syntax Specifications: ABNF,” October 2005.) of the ABNF specification). Note that it is the white space characters at the beginning of the subsequent lines (including the commented ones) that conspire to extend the reach of those rules across several lines.

7.  Limitations of xml2rfc

• The figure element's title attribute is ignored, except when generating HTML.
• The xref element's pageno attribute is ignored.

8. References

Appendix A.  MacOS 9 Installation (courtesy of Ned Freed)

1. Install Tcl/Tk 8.3.4
2. When you performed Step 1, a folder in your Extensions folder called Tool Command Language was created. Create a new folder under Extensions, with any name you like.
3. Drag the file xml2rfc.tcl onto the Drag & Drop Tclets application that was installed in Step 1.
4. When asked for an appropriate wish stub, select Wish 8.3.4.
5. The Drag & Drop Tclets application will write out an executable version of xml2rfc.

Appendix B.  rfc2629.xslt (courtesy of Julian Reschke)

The file rfc2629.xslt can be used with an XSLT-capable formatter (e.g., Saxon, Xalan, xsltproc, and most browsers) to produce HTML. A word of warning though: the XSLT script only supports a limited subset of the processing instruction directives discussed earlier (Processing Instructions). The latest version (and full distribution ZIP file) can be downloaded from the original site which also hosts its documentation.

Appendix C.  MS-Windows XP/Cygwin Installation (courtesy of Joe Touch)

1. install Cygwin: follow instructions at the Cygwin website (also visit the Cygwin pages on the Tcl Wiki), make sure to select tcltk in Libs
2. place a copy of xml2rfc files on a local drive, e.g., in C:\xml2rfc NOTE: for xml2rfc-1.26 and earlier, see NOTE below.
3. place a copy of bibxml* files on a local drive, e.g., in C:\xmlbib\
4. edit .xml2rfc.rc to indicate the bibxml* library path, e.g., as per Step 3, change ~/rfca/bibxml/* to /cygdrive/c/xmlbib/*
5. run xml2rfc as follows: tclsh /cygdrive/c/xml2rfc/xml2rfc.tcl

NOTE: for xml2rfc-1.26 and earlier ONLY, add an additional modification in Step 3:

Patch .xml2rfc.rc (xml2rfc-1.26 and earlier). The purpose of the patch is to append library names in a format compatible with the OS; on MS-Windows XP, this replaces the Cygwin's / with MS-Windows' \.

--- .xml2rfc.rc.orig    Thu Jul 24 13:58:00 2003
+++ .xml2rfc.rc    Wed Oct 20 10:59:02 2004
@@ -9,7 +9,8 @@
 if {[catch { set env(XML_LIBRARY) } library]} {
     set library ""
     foreach bibxmlD [lsort -dictionary [glob -nocomplain ~/rfcs/bibxml/*]] {
-        append library $sep$bibxmlD
+        set natbibD [file nativename $bibxmlD]
+        append library $sep$natbibD
     }
 }

Appendix D.  A Special Thanks

A special thanks to Charles Levert for the v1.29 release, which includes many internal improvements made to the rendering engines.

Appendix E.  Copyrights

Copyright © 2003–2008 Marshall T. Rose

Hold harmless the author, and any lawful use is allowed.

Authors' Addresses