PDF Generation Progress: Now With OASIS Styling

[Eliot Kimber <ekimber@reallysi.com>]

Thanks to Suite Solutions, I've integrated their OASIS styling
customizations for the PDF2 transform and verified that they work.

I've committed to the SVN repository a Toolkit plugin that packages the
customizations and provides a placeholder for any plugin-provided XSLT
extensions we might need.

I've created a set of Ant build scripts that automate the deployment of the
plugin to your local Toolkit instance and manage the generation of the PDF
for the ArchSpec, LangRef, and full spec.

Note that to render the language reference or the full spec you'll to fix
the FO plugin build.xml file to provide enough memory and you may require a
64-bit JVM to have enough memory available (I'm not sure what the actual
requirement is--I haven't had a chance experiment and see what it actually
requires). But the standalone ArchSpec should render OK.

The Ant tasks use a property to know where the Toolkit is, which you can set
by creating a file named "build.properties" or ".build.properties" in your
home directory that sets the property "dita-ot-dir" to the appropriate
location, e.g., my .build.properties file has this entry:

dita-ot-dir=${oxygen_home}/frameworks/dita/DITA-OT

Since I use a Toolkit integrated with OxygenXML.

The scripts are set up to use Antenna House by default, so you also need to
set the axf.path property if you use Antenna House, otherwise change value
of the pdf.formatter property to "xep" or "fop", as needed (not sure FOP can
actually handle the rendering, haven't tried since I have both XSL Formatter
and XEP available):

axf.path=c:/Program Files/XSL Formatter

Or whatever your path should be. You can set any properties in
build.properties and they'll override any settings in the Ant scripts.

Once you create that file you should be able to start a Toolkit command
window using startup.sh or startup.cmd and run the Ant command, move to the
root of your DITA spec working copy, and issue this Ant command:

ant -f build_dita_spec_base.xml

That should copy the spec-specific Toolkit plugins to your local Toolkit
instance and integrate them.

You should then be able to run the different PDF generation Ant scripts,
e.g.:

ant -f build_archspec_pdf.xml

Note that the Toolkit startup scripts set the memory for the JVM that runs
the Ant tasks to 512Meg, which is definitely not enough to process the
langref and full spec. To change it, you can either modify the startup
script to change the value of the ANT_OPTS environment variable or you can
reset the ANT_OPTS environment variable after you start the command window,
e.g.:

export ANT_OPTS='-Xmx1000m
-Djavax.xml.transform.TransformerFactory=net.sf.saxon.TransformerFactoryImpl
'

That is, just copying the existing value of ANT_OPTS and changing the -Xmx
parameter to a higher value. If you have a 64-bit machine, you can add
"-d64" to ANT_OPTS to turn on 64-bit mode and then you can use values larger
than 2Gig (or whatever the 32-bit limit is on your machine, which will
typically be between 1.2 and 1.8Gig, depending).

The build scripts have been set up to allow adding support for other output
types than PDF but I haven't set those up.

I was hoping I'd be able to provide a Web-accessible production server for
use by the TC editors for on-demand processing of the spec, but RSI doesn't
have any machines available and we can't really afford to set up a dedicated
machine just for this purpose. But if someone had a net-accessible machine
with at least 4Gig of RAM it would be pretty easy to set up a trivial Web
server to do on-demand production of the spec, since the source can be
fetched from the SVN repository as needed and rest is just running some Ant
scripts.

Cheers,

Eliot

-- 
Eliot Kimber
Senior Solutions Architect
"Bringing Strategy, Content, and Technology Together"
Main: 610.631.6770
www.reallysi.com
www.rsuitecms.com