Skip to content
Go to file

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

XProc 3.0: An XML Pipeline Language

This is the home of the XProc 3.0 language specification developed by the XProc next community group. The specifications for the step libraries are maintained in the step repository.

Drafts are published automatically at


The XProc community is using GitHub to manage the development of this specification. Please pull the repository, make improvements, and propose changes in the form of pull requests.

Continuous integration

The XProc specification is built automatically with Travis CI.

To build and publish the spec on your gh-pages, setup the gh-pages branch, configure Travis CI to run for your repo, and then create the following secure environment variables for your repo in the Travis CI Settings page for your fork:

  • GH_TOKEN="your git token"
  • GIT_EMAIL=""
  • GIT_NAME="Your Name"
  • GIT_PUB_REPO="you/3.0-specification"

The GIT_TOKEN must be a personal access token. The GIT_PUB_REPO must be the repository where you wish to publish the results. The publications scripts will push the published documents to the gh-pages branch.

Travis CI will then publish your changes everytime you do a commit to your master branch. Travis CI cannot publish gh-pages for pull requests.

How it works

The documents are built by a gradle task that runs XML Calabash and other tools. The individual specifications are in sub-projects. At the time of this writing, there are two specifications in this repository:

  • overview is the overview page that points to all the specs
  • xproc is XProc 3.0: A Pipeline Language

The default gradle target, allspecs, will build all of the specifications. The built specifications are in the build/dist/ directory or directories below it.

Within each subproject, the “source” for each specification is in src/main/xml/specification.xml. There may be other files as well. Use XInclude to break specs into pieces if you wish.

About the workflow

It’s all a bit complicated. These are some notes.

  • The published specs are the gh-pages branch of this repository.

  • Three repositories update the gh-pages branch: this repository publishes the language specification; the 3.0-steps repository publishes the steps; the 3.0-grammar repository publishes the grammar.

  • Publishing the grammar relies on the steps! So 3.0-grammar should be rebuilt whenever the steps change and this repos should be rebuilt whenever the core grammar changes!

  • All specifications have a glossary; if the glossary turns out to be empty, because there are no term definitions (firstterm elements) in a specification, it will be elided automatically.

<xi:include href="../../../build/glossary.xml">
    <glossary xml:id="glossary">
      <para>Glossary needs to be generated</para>
  • Similarly, for steps, there is a step errors appendix. It will be populated automatically and elided if there are no step errors.
<appendix xml:id="app.step-errors">
<title>Step Errors</title>
<para>The following <glossterm baseform="dynamic-error">dynamic errors</glossterm>
can be raised by the <code>p:validate-with-relax-ng</code> step:</para>
  • There is a single, master bibliography in src/main/xml/bibliography.xml. In each specification, use biblioref to refer to bibliography entries. Create a bibliolist with empty bibliomisc elements to pull in the relevant entries from the master bibliography:
<appendix xml:id="references">
<bibliomixed xml:id="xproc30"/>
<bibliomixed xml:id="xproc30-steps"/>
<bibliomixed xml:id="iso19757-2"/>
<bibliomixed xml:id="relaxng-compact-syntax"/>
<bibliomixed xml:id="relaxng-dtd-compat"/>
  • Examples and graphics are also a little complicated. I’ll describe them here if anyone asks. :-)

The build process is owned by norm; bug him if you have difficulties.


A community-driven effort to define an XProc 3.0 specification (formerly 1.1)



No releases published
You can’t perform that action at this time.