10. 2.000beta8

Enhancements | Fixes | Incompatibilities

10.1. Enhancements

There are lots of new things in this version. The general enhancements are:

The table enhancements are:

The page formatting enhancements are:

The configuration enhancements are:

Details are given below.

Note: For html and txt targets, this release is slightly faster than the previous one. For mif (and ps), this release may be slightly slower (as SDF templates, not FrameMaker ones, are now used).

10.1.1. Easier enumerated values

Enumerated values no longer need to be quoted. In a set of name-value pairs, if an expression is a single word where:

  • the first character is uppercase
  • the remaining characters are lowercase

then the word is assumed to be an enemerated value and it is automatically converted to a string. For example, the following two lines are now equivalent:

   P1[align="Center"] My Title
   P1[align=Center] My Title

10.1.2. Look improvements

The overall appearance (i.e. look) of an SDF document can be controlled via the -k option or the OPT_LOOK variable. SDF now supports the following looks:

  • simple - the default look, useful for general documentation
  • fancy - QSD-like look, but without any logos
  • infomap - look based on Information Mapping™
  • overhead - look suitable for overhead transparencies.

Note: For backwards compatibility, plain is now an alias for simple.

10.1.3. Style rationalisation

SDF now supports the styles (i.e. document types) below.

Style Purpose
document a normal document
manual a manual
paper a technical paper
admin generic administration document
fax a fascimile
memo a memorandum
newslttr a newsletter
minutes minutes of a meeting
listing a source code listing

The build_title macro is used to build the cover page for the general styles. The title filter is used to build the title block for the administation styles. The miscellaneous styles do not have a title.

Adding new styles for a look is now much easier than it was previously. Furthermore, manual is now just another style rather than a special mode.

10.1.4. More build_title control variables

build_title now supports the following variable:

  • DOC_COVER - the logical name of the cover page to use.

Supported DOC_COVER values and their meanings are given below:

Value Meaning
original SDF's original cover page layout
project a cover page useful for project documentation
manual the cover page used when OPT_STYLE is manual
paper the cover page used when OPT_STYLE is paper

The manual cover supports a new macro called DOC_OFFICES, i.e. if DOC_OFFICES is defined, its contents are appended after the inside cover page.

10.1.5. Miscellaneous things

The SDF What's Planned document and bugs.html have been replaced by the SDF Bug Database. In fact, the organisation of the documentation has been improved quite a bit.

Any document can now be converted to a FrameMaker book, i.e. it is no longer necessary to specially organise files in order to create a FrameMaker book.

Tables of style columns (the default) now have a thin line above group rows.

Tables can now be specified as list items by using the listitem parameter of the table filter. The value is the logical indent of the list (e.g. 1, 2, etc.). When generating HTML which contains a table within a numbered list, the listitem parameter is necessary in order to prevent the numbering from resetting.

The new box filter can be used to place a box around one or more lines of text.

The new center filter can be used to center a region of text.

The new paragraph attribute called wide can be used to specify that a paragraph straddles the side-head area in addition to the main text area. Likewise. class filters (e.g. terms, references, etc.) now also support the wide attribute.

By default, SDF automatically generates an id attribute for all headings in the table of contents. However, it is occasionally necessary to disable this, particularly when the same heading appears multiple times in a file. The new paragraph attribute noid can be used to prevent an id being generated for a paragraph.

The title for an administration-style document can now be changed by either setting the DOC_TYPE variable or by using the new type parameter to the title filter. For example, if you want the title "Urgent Memo" rather than the default "MEMORANDUM" for a memo, you can now do the following:

!init OPT_STYLE="memo"
!block title; type="Urgent Memo"
usual name-values pairs go here ...

The title filter now also supports a format parameter which can be used to tune the column widths, when necessary.

RTF can now be generated via FrameMaker by using the new sdf2fmrtf alias, i.e. the command is (say):

   sdf -2fmrtf myfile.sdf

When RTF is generated via FrameMaker, Word's standard style names are used, where possible. This makes it easier to import the RTF file into an existing Word document.

Lists of tables and lists of figures can now be generated for normal documents, i.e. it is no longer necessary to use a FrameMaker book to build these lists.

The new variable TXT_MARGIN can be used to control the width for txt output. Likewise, the new variable POD_MARGIN can be used to control the width of tables for pod format.

The new bugtrack module can be used to build a simple bug tracking system like SDF's bug database.

The new namevalues macro can be used to output a set of name-value pairs for an object. The parameters are:

  • the class name
  • the object name
  • a comma-separated list of attribute names.

For example:

!namevalues 'bugs'; 'sd0001'; 'Priority,Status,Type'

Finer control over object catalog formatting is now available via the new style and headings attributes. style is the table style to use. The default value is plain. If headings is set, column headings are output. (By default, they are not.)

10.1.6. Easier column widths

SDF now supports dynamic column widths for paper documentation, i.e. if a width is not specified for a column, then the column is sized based on the text within it and the space available. Proportional column widths are also supported.

The format attribute of the table filter is either:

  • a single number, in which case each digit represents 10% of the width available to the table, or
  • a comma-separated list of column width specifications.

Examples of the column width specifications now supported are given below.

Value Meaning
30pt an exact size (other supported units are cm, mm, " and in)
30% a percentage of the size available
30 a percentage of the size available (% is implicit)
10-20 dynamic size between 10% and 20% of the total width
-20 dynamic size between 0% and 20% of the total width
10- dynamic size between 10% and 100% of the total width
- dynamic size between 0% and 100% of the total width
3* 3 units of the remaining space
* same as 1*

For example, in the table below, the second column will be twice the size of the last column.

!block  table; format="20,2*,10,*"
Name    Column2             Column3     Column4
A       B                   C           D
X       Hello dear world    Y           Z

The output is:

Name Column2 Column3 Column4
X Hello dear world Y Z

If a column is not given a size, the following rules are used:

  1. The last unspecified column size is implicitly '*' (i.e. the rest), unless the narrow attribute is set, in which case the size is implicitly '-' (i.e. as much as needed).
  2. The other unknown sizes are implicitly '-'.

For example, the first and third columns in the table below will be dynamically sized. The first column will take as much space as required and the last column will expand so that the table takes the full width of the text area.

!block  table; format=",30,,10"
Name    Column2             Column3     Column4
A       B                   C           D
X       Hello dear world    Y           Z

The output is:

Name Column2 Column3 Column4
X Hello dear world Y Z

However, in the example below, each column will only take as much space is required, making the table narrower than it would be otherwise.

!block  table; narrow
Name    Column2             Column3     Column4
A       B                   C           D
X       Hello dear world    Y           Z

The output is:

Name Column2 Column3 Column4
X Hello dear world Y Z

If an = character is used in place of a - character for a column width, then those columns will be equalised in size. For example, the second and forth columns in the table below will be made equal in size.

!block  table; format="20,5=30,10,="
Name    Column2             Column3     Column4
A       B                   C           D
X       Hello dear world    Y           Z

The output is:

Name Column2 Column3 Column4
X Hello dear world Y Z

Note: As previously, the format parameter has no impact on HTML generation, i.e. all columns in HTML remain dynamically sized. Dynamic sizing is also ignored for txt and pod targets.

10.1.7. Heading/footing rows for tables

The table filter now supports the following attributes:

  • headings - the number of heading rows at the top of the table
  • footings - the number of footing rows at the end of the table.

If the headings attribute is not defined, then as previously done, the column headings are generated using the column names given on the parsing line. For example, the column headings in the table below will be Name and Age:

!block table
Name    Age
Bill    42

Alternatively, if the headings attribute is defined, then that number of data rows are used as the column headings, i.e. the parsing line is not used to build the column headings. For example, the column headings in the table below will be "Title" and Age (remembering that a filter attribute is implicitly given the value 1 is no value is supplied).

!block table; headings
A       B
"Title" Age
Bill    42
Sally   23

Likewise, the column headings below will be Preferred Title and Likely Age with each heading taking 2 rows.

!block table; headings=2
A           B
Preferred   Likely
Title       Age
Bill        42
Sally       23

Collectively, these new attributes improve things in several ways:

  • heading rows can now include non-alphanumeric characters
  • macros (e.g. change bars) can be applied to heading rows
  • multi-line heading and footing rows are now supported.

10.1.8. Table positioning

The horizontal alignment and vertical placement of a table can be controlled by setting the align and placement parameters of the table filter respectively. The permitted align values are:

Value Meaning
Left left-align the table
Center center the table
Right right-align the table
Inner align the table with the inner margin
Outer align the table with the outer margin

Note: The wide parameter changes the left indent of a table to include the sidehead of a page. Therefore, the wide parameter will impact the horizontal positioning of any table which is not right-aligned.

The permitted placement values are:

Value Meaning
Float next column if necessary
Pagetop top of the next page
Columntop top of the next column
Lefttop top of the next left-hand page
Righttop top of the next right-hand page

10.1.9. Table filtering and sorting

Tables can now be filtered and sorted by using the new where and sort attributes of the table filter. These attributes are also supported by the class filters (e.g. terms, references). In either case, filtering is done before sorting.

The where attribute takes an expression which is evaluated for each record. Special symbols available are:

Symbol Meaning
$_ the current record
$o{"xyz"} the value of column xyz

sort takes a comma-separated list of column names to sort on. If no columns are specified, the data is sorted using all columns in the order in which they appear. All sorting is done alphabetically - numeric sorting is not supported.

10.1.10. Landscape tables

Landscape tables are now supported via the landscape parameter of the table filter. The value is the height allocated to the area in which the table is placed. If a unit is not specified, the value is assumed to be a percentage of the text column height. For convenience, a value of 1 implies a full page table. Some examples are given below.

Value Meaning
landscape="50pt" height allocated to table is 50 points
landscape="50%" half page table
landscape=50 half page table (% is the default units)
landscape=1 full page table (1 implies 100%)
landscape full page table (syntactic shorthand for above)

Note: This feature is currently use at your own risk. In particular, long tables and table titles confuse it badly. Furthermore, the align and placement parameters are effectively ignored for landscape tables.

10.1.11. Column alignment

The colaligns parameter of the table filter can now be used to control the alignment of text within columns of a table. For example:

!block  table; colaligns="LCCR"
Name    Column2             Column3     Column4
A       B                   C           1.0
X       Hello dear world    Y           10.2

The output is:

Name Column2 Column3 Column4
A B C 1.0
X Hello dear world Y 10.2

The value of colaligns is usually a sequence of the letters L, C and R (which mean what one would expect). If you prefer, a comma-separated list of the values Left, Center and Right can be specified. For example:

!block  table; colaligns="Left,Center,Center,Right"
Name    Column2             Column3     Column4
A       B                   C           1.0
X       Hello dear world    Y           10.2

10.1.12. Cell attributes

Cells within a table can now be given attributes by preceding the cell value with a semicolon-separated list of name-value pairs enclosed in square brackets. For example:

!block  table; colaligns="LCCR"
Name    Column2             Column3     Column4
A       [align=Left]B       C           [bgcolor=Green]1.0
X       Hello dear world    Y           [bgcolor=Red]10.2

The output is:

Name Column2 Column3 Column4
A B C 1.0
X Hello dear world Y 10.2

The cell attributes supported are given below.

Name Value
align horizontal alignment (Left, Center, Right)
valign vertical alignment (Top, Middle, Bottom, Baseline)
cols the number of columns this cell spans (default is 1)
rows the number of rows this cell spans (default is 1)
bgcolor background colour of cell (see below)
PS only:  
fill background colour fill percentage
truling ruling setting for top of cell
bruling ruling setting for bottom of cell
lruling ruling setting for left of cell
rruling ruling setting for right of cell
angle angle of text (0, 90, 180, 270)
HTML only:  
nowrap disable word wrap for this cell
sdf treat the cell text as SDF (rather than as phrase text)
tag phrase tag to apply to cell text
paratag paragraph style to apply to cell text

For PS (i.e. MIF) generation, the supported colour values are Black, White, Red, Green, Blue, Yellow, Cyan and Magenta. If a different colour is specified, it is ignored. The supported fill values are 100, 90, 70, 50, 30, 10 and 3. If a fill value is not specified, 100% fill is used.

For HTML generation, any of the HTML colours names (including those supported for PS generation) or the "#rrggbb" form can be used.

The permitted ruling values are Vthin, Thin, Medium, Thick and Double.

The sdf, tag and paratag attributes control the way in which the cell text is converted to SDF:

  1. If sdf is set, the cell text is already SDF.
  2. Otherwise if tag is set, the SDF paragraph is paratag:{{tag:text}}.
  3. Otherwise, the paragraph is paratag:text.

tag is usually set via the tags or groups parameters of the table filter.

Note: paratag is not yet implemented.

10.1.13. Spreadsheet calculations

The spreadsheet expression evaluator and the documentation below was written by Tim Hudson (

Spreadsheet style calculations have been introduced into SDF using the standard [[ ]] syntax with a prefix of + (or =) indicating that the expression is to be evaluated by the calculation routines.

This extension has been loosely modelled on Microsoft Excel® in terms of the initial functions supported and the syntax used.

Note: Calculation support for a table can be disabled by adding in an attribute of nocalcs (otherwise the pointers required to table data that are needed when doing spreadsheet calculations occur for each table cell).

Each cell in a table has an cellid which is made up of a single uppercase letter indicating the column index and a number indicating the row index (counting from 1 and excluding the heading rows). The upper left cell is hence A1.

An example grid indicating cellids:

Title1 Title2 Title3 Title4 Title5
A1 B1 C1 D1 E1
A2 B2 C2 D2 E2
A3 B3 C3 D3 E3
... ... ... ... ...
A100 B100 C100 D100 E100

A range of cellids is specified using the syntax cellid1:cellid2. For example: A1:C1 is exactly the same as A1,B1,C1

An expression consists of a combination of standard Perl operators and spreadsheet functions and cellids or cellid ranges.

Standard Perl operators include:

  • + - * /

Spreadsheet functions use the syntax FUNCTION(ARG1,ARG2,...ARGN).

The following functions are supported:

  • AVERAGE - the average - SUM(ARGS)/COUNT(ARGS)
  • SUM - the sum of the args - same as ARG1+ARG2+...+ARGN
  • MIN - the minumum argument value
  • MAX - the maximum argument value
  • COUNT - the number of arguments
  • PRODUCT - the product of the args - same as ARG1*ARGN*...*ARGN
  • ROWSUM - the SUM of all the cells in the row to the left of the current cell
  • ROWPROD - the PRODUCT of all the cells in the row to the left of the current cell
  • COLSUM - the SUM of all the cells in the column above the current cell
  • COLPROD - the PRODUCT of all the cells in the column above the current cell

A simple example:

!block table; style="grid"
Count         Price        Total
10            5            [\[=A1*B1]\]
15            5.23         [\[=ROWPROD]\]
[\[=COLSUM]]   [\[=B1+B2]\]   [\[=COLSUM]\]

Which generates the result below. (Ok, summing two prices is meaningless, but it illustrates the syntax.)

Count Price Total
10 5 50.00
15 5.23 78.45
25.00 10.23 128.45

Note: values are available until the next table is processed so you can refer to data inside normal paragraphs after the table like this [\[=A1]\] which evaluates to 10.00

A spreadsheet expression will recursively evaluate any expressions contained in cells that are used in an expression. In the example above, the expression in cell C3 depends on the results of the expression in cell C1 and C2.

10.1.14. Arbitrary page sizes

Generally speaking, SDF can now support arbitrary page sizes. The catch is that the page size must now be specified up front via the OPT_PAGE_SIZE variable on the init line or via sdf's -S option. (Previously, the (no longer supported) DOC_PAGE_SIZE variable could be specified anywhere in the document.)

The supported values of OPT_PAGE_SIZE are given below. The default page size is global.

Name Width Height Comment
global 21.0cm 11.0in will fit on either A4 or letter
A3 29.7cm 42.0cm  
A4 21.0cm 29.7cm  
A5 14.8cm 21.0cm  
B4 25.7cm 36.4cm  
B5 17.6cm 25.0cm  
letter 8.5in 11.0in  
legal 8.5in 14.0in  
tabloid 11.0in 17.0in  

Note: global is 0.23" narrower than letter and 1.76cm shorter than A4.

New page sizes can be defined in sdf.ini. A rotated version is also available for each page size. Rotated sizes are named with an appended R. For example, the rotated A4 size is called A4R.

10.1.15. Headers and footers

SDF now provides 3 ways of controlling headers and footers:

  • The OPT_HEADINGS variable (high level control)
  • PAGE_page_HF_ICOn variables (medium level control)
  • PAGE_page_HF macros (low level control)


  • page is FIRST, RIGHT or LEFT
  • n is the line number within the header or footer (usually 1 or 2).

The meaning of each page type is explained below.

Page Usage
FIRST the first page in the document
RIGHT the page used for single-sided documents and the right hand side of double-sided documents
LEFT the left hand side of double-sided documents; ignored for single-sided documents

Supported OPT_HEADINGS values and their meanings are given below:

Value Meaning
0 no header, no footer
1 single line header, single line footer
2 two line header and two line footer
3 two line header and three line footer
4 two line header and four line footer

10.1.16. Header/footer special tags

The following special tags are supported within headers and footers:

Tag Meaning
PAGENUM current page number
PAGECOUNT highest page number
PARATEXT paragraph text (e.g. Hardware requirements)
PARANUM paragraph number (e.g. Appendix A)
PARANUMONLY paragraph number only (e.g. A)
PARASHORT value of short attribute of paragraph
PARALAST text of last paragraph on page with the nominated style

The PARA* tags require a comma separated list of paragraph styles to be nominated as the text of the phrase, e.g. {{PARATEXT:H1,A1,P1}}.

The PARASHORT tag is useful for placing an alternative heading in a header, say. For example, you might set up your header to include {{PARASHORT:H1}} and have the following text in your document:

  H1[short='Getting started'] Getting started with SDF

The PARALAST tag is useful for producing dictionary-like headers.

10.1.17. Header/footer borders

High level control over header/footer borders is provided by the OPT_BORDERS variable. The supported values and their meanings are given below:

Value Meaning
0 no header/footer borders
1 line below header, line above footer
2 lines above and below header, line above footer

Finer control is available by setting the following variables:

Name Value
PAGE_page_HEADER_BORDER border specification string
PAGE_page_FOOTER_BORDER border specification string

where page is either FIRST, RIGHT or LEFT.

A border specification string is a comma-separated list of attributes which collectively describe the border. The format of each attribute is name[=value]. The supported attributes are:

  • top - a line above the area
  • bottom - a line below the area
  • box - a box around the area
  • radius - for a box, the radius of the corner.

For top, bottom and box, the value of the attribute is the line width in points.

10.1.18. Page backgrounds

In order to support fancy background images, SDF supports the following variables:

Name Value
PAGE_page_BACKGROUND the logical image name

where page is either FIRST, RIGHT or LEFT.

When FrameMaker is being used to generate PostScript, the image name is mapped to a master page called backgrnd within a file called backgrnd.mif.

Note: At the moment, objects from the master page (excluding TextRects) are directly transferred to the generated MIF file. This means that objects in the lower right hand corner of an A4 master page will not will be positioned nicely if the paper size is changed to A5, say.

10.1.19. Page layout variables

Page margins can now be controlled via the following variables:

Name Value
OPT_MARGIN_TOP size of the margin above the header, if any
OPT_MARGIN_BOTTOM size of the margin below the footer, if any
OPT_MARGIN_INNER size of the inner margin
OPT_MARGIN_OUTER size of the outer margin

The positioning of headers and footers can now be controlled via the following variables:

Name Value
PAGE_page_HEADER_HEIGHT height of the area allocated to the header
PAGE_page_HEADER_GAP size of the gap between the header area and the text area
PAGE_page_FOOTER_HEIGHT height of the area allocated to the footer
PAGE_page_FOOTER_GAP size of the gap between the text area and the footer area

where page is either FIRST, RIGHT or LEFT.

After the margins and header/footer are allocated, the remaining text area contains 1 or more text columns and an optional side head. The layout of these is controlled by the following variables:

Name Value
OPT_COLUMNS numbers of text columns
OPT_COLUMN_GAP space between columns
OPT_SIDEHEAD_WIDTH size allocated to the sidehead
OPT_SIDEHEAD_GAP space between the side-head and first text column

The following read-only variables provide information on the page layout:

Name Value
DOC_PAGE_WIDTH width of page
DOC_PAGE_HEIGHT height of page
DOC_FULL_WIDTH width of text area including the side-head
DOC_TEXT_WIDTH width of text area available for text columns
DOC_TEXT_HEIGHT height of text area
DOC_COLUMN_WIDTH width of a text column

All the information variables are in units of points. The text area is the area allocated to text, excluding the header and footer (if any) on the RIGHT page.

10.1.20. Page control within a book

When a book is being generated, each component of the book may have its own FIRST, RIGHT and LEFT pages, i.e. all of the macros and variables above support an extended set of page values component_FRL where:

  • component is the name of the component

The component names supported are given below.

Name Meaning
FRONT the cover component
PRETOC components before the contents
TOC the table of contents
LOF the list of figures
LOT the list of tables
PRECHAPTER components before the chapters
CHAPTER a normal chapter
APPENDIX an appendix
PREIX components before the index
IX the index

If a component does not have a component-specific macro/variable defined, then the generic macro/variable is used.

10.1.21. FrameMaker templates are no longer required

Previously, SDF produced MIF documents by merging generated MIF paragraphs into FrameMaker templates. This approach was easy to implement but maintaining the FrameMaker templates was difficult as every look required several templates for each page size!

FrameMaker templates can still be used but are no longer needed, nor recommended. Instead, the definitions of paragraph styles, phrase styles, table styles and graphic (border) frames is now done via SDF templates. SDF templates have the following advantages:

  1. Templates can be parameterised so that a single template can support any number of page sizes.
  2. Each look only requires a single template, rather than a template for each component of a FrameMaker book.
  3. Templates can inherit definitions from other templates, making it much easier to create and maintain templates.
  4. Each definition supports inheritance, so a new paragraph style, say, can be defined in terms of the differences between it and a parent style.

The new sdtgen command can be used to build an SDF template from a FrameMaker one. Typically, sdtgen is used to create an initial template which is then simplified via definition inheritance and template inheritance.

10.1.22. Event processing enhancements

The following new symbols are now available within event processing for paragraphs:

Symbol Meaning
$level the current heading level (before this paragraph)
$prev_style the style of the previous paragraph
$prev_text the text of the previous paragraph
%prev_attr the attributes of the previous paragraph

These symbols make it much easier to do some really nice things. For example, the following rule can be used to put level 2 headings on a new page unless the heading is the first one within a chapter:

!on paragraph '[HAP]2';; if ($level != 1) {$attr{'top'} = 1}

10.2. Fixes

Tim Hudson has contributed a number of fixes to this release including:

  • this version should now work with Perl 5 again - the previous version triggered a Perl 5 syntax error
  • the IMPORT pragma now supports filename extension adding ala the import macro
  • centred figures now work as expected in HTML
  • numerous Windows help generation fixes/enhancements.

Other fixes include:

  • the renaming of to is now done by sdfbatch, so failures (due to lack of disk space, say) are now handled much better than they previously were
  • topics mode now generates an error if the heading of a topic is duplicated within a file (see Miscellaneous things above)
  • headings containing a ? character should no longer cause problems
  • some bugs in sdfget's extraction of documentation from C code have been fixed (thanks to Keith Ponting).

10.3. Incompatibilities

As the default page size is now global rather than A4, page breaks may occur slightly earlier than they previously did. The advantage of changing to global is that documents should now print successfully on US letter paper.

Memos, faxes, minutes and newsletters no longer have the company address in the top right hand corner. (I'll fix this soon, but it is not a high priority at the moment.)

As txt and pod outputs use a different algorithm for deciding column widths when a table does not have a format parameter, some table cells may no longer fit when these output formats are used.

The table attribute previously called headings has been renamed to parseline to better reflect its purpose and to make way for the new meaning.

The DOC_TOCTEXT variable has been renamed to DOC_TOC_TITLE. The DOC_TOCGRAPHIC variable has been renamed to DOC_TOC_GRAPHIC.

The newsletter style has been renamed to newslttr.

overhead is now a look, rather than a style. The overhead look is untested and rarely used, so use it at your own risk.

The aliases sdf2htmls and sdf2txts have been renamed to sdf2fmhtml and sdf2fmtxt respectively.

/usr/local/bin/perl5 is now the default perl interpreter used.