This is the permanent status page for Calabash. Calabash is an implementation of XProc: An XML Pipeline Processor, a specification being developed by the W3C to address questions about the XML processing model. Version 1.0.15 is now available.

XML Calabash LogoCalabash runs XProc pipelines. It is a command-line application. This version of Calabash is supposed to implement the 11 May 2010 Recommendation of the XProc specification.

The current release is 1.0.15 from 28 October 2013. This is a stable release.

Recent changes are now recorded on the download page.

Calabash's performance against the test suite is reported. (Most releases so far have passed most, if not quite all, of the tests.)

Calabash is implemented in Java and should run on any platform that supports Java 1.5 or later and has a command-line. You can get “calabash” from the download page on It is available under the terms of the GNU General Public License, Version 2.0.

Previous releases:

Calabash is maintained by Norman Walsh. Please report any bugs that you encounter.

If you'd like to build the project yourself, you can browse the Git repository or access it directly from

Calabash was first announced on 24 August 2008. It is a complete rewrite of an earlier project named simply The XML Pipeline Processor.


There doesn't seem to be any source code at

Posted by Gavin Carothers on 03 Sep 2008 @ 05:17pm UTC #

It was there, you just couldn't see it :-) I've fixed that now.

Posted by Norman Walsh on 05 Sep 2008 @ 11:43am UTC #

Thanks for your work on this.

I'm going to play with it a bit this week.

Posted by Chris on 30 Sep 2008 @ 04:54pm UTC #

The fact that you keep on adding dependancies on foreign proprietary software shows how XML has failed. Obviously step #1 of xproc support work should be to have open-source implementations of all the XML Standards needed for xproc itself available. I like the idea of xproc; I hate the idea of having to throw out hundreds, if not in sum thousands of dollars for licenses just to run the full xproc vocabulary. E.g. I have the perfect usage scenario for xproc. validate query validate transform split upload etc. Seems to start off with xproc seriously, I first gotta implement the pack of xml standards. Wake up W3C! Where are YOUR prototypes?!

Posted by Martin Weber on 27 Oct 2008 @ 05:05pm UTC #


The only standard steps in Calabash that require any proprietary software are p:validate-with-xml-schema and p:xsl-formatter.

In each case, it's simply a matter of expediency on my part. I will certainly support Xerces for schema validation eventually, and probably FOP for XSL formatting.

All of the other dependencies are on open source packages of one flavor or another.

Posted by Norman Walsh on 30 Oct 2008 @ 01:38pm UTC #

Norman, mean "so long as you don't try to use the two year old XPath2 / XSLT2 standards" which again forces you to have DTDs available which again .. or you go down the money route. That's what I'm talking about. Anyways, I'd like to seriously test this but with saxon as a requirement I'll probably have to hack some first. Let's see what my employer says..

Posted by Martin Weber on 04 Nov 2008 @ 06:50pm UTC #


There's nothing commercial about the Saxon support except for the p:validate-with-xml-schema step, which I've already explained is a question of implementation convenience, not any sort of refusal to implement an entirely free suite of steps.

Posted by Norman Walsh on 07 Nov 2008 @ 02:07am UTC #

Regarding XPath2 / XSLT2: do either of these requires DTDs or any other form of validation? I thought you just don't get a lot of the schema-aware goodness if you don't have validation. There are, however, lots of other reasons why these versions are a big step forward from the old ones.

By the way, is there an open-source schema-aware XSLT2 processor available? I'm fully aware of the Saxon-B/SA split.

Posted by Ed Davies on 16 Nov 2008 @ 11:52am UTC #


You don't need DTDs or XML Schemas to use XPath 2.0 or XSLT 2.0. You can use DTDs, of course, and you can use XML Schema if you have a schema-aware processor. I'm not aware of any open-source, schema-aware XSLT 2.0 processors.

Personally, I stick with RELAX NG validation and don't care about schema-awareness at the XSLT level (beyond simple types that are available in a basic processor).

Posted by Norman Walsh on 17 Nov 2008 @ 02:42pm UTC #

I appreciate the nature of this Calabash project and the W3C work. It is on a valuable track.

We've been working on a much narrower development project on a homebrew basis, but it does reflect the qualities of creating a modularized pipeline of XSLT transforms. It uses an ASP script controlling things together with MSXML objects and methods (on a Windows server obviously). It's oriented to simply grabbing static XML/HTML inputs, running a specified series of transforms, then displaying the result in a browser as XHTML, forms, slightly animated at that point by embedded Javascript (nothing fancy at all, no frills).

Our framework simply leverages naming the inputs and XSLTs on the URL http call, e.g.,Transform1,Transform2,ShowTable

Thank you Norman!

Posted by Dave Porter on 02 Dec 2008 @ 06:44pm UTC #

What about auto-discovery of configuration.xml-like files on the classpath, akin to JAR service lookup?

This would mean that extension steps can be supported by Calabash just by dropping a JAR with the implementation and extra configuration file onto Calabash's classpath.

Such extra configuration files should probably be able to override the built-in <implementation> declarations, in order to, for instance, implement <p:xsl-formatter> with FOP.

Posted by Alex Hvostov on 17 Jan 2009 @ 12:12am UTC #

Can you tell me how to use p:xsl-formatter step. I am getting some errors when I use this step.

I am running calabash with the following options. xpl\test-pdf.xpl

xep is added in the classpath.

Below is the error message:

Error : file:/C:/Sesi/personal/xml/xproc/calabash-0.9.14/calabash-0.9.14/xpl/test-pdf.xpl:2: Output port 'result' on anonymous step declare-step at file:/C:/Sesi/personal/xml/xproc/calabash-0.9.14/ca labash-0.9.14/xpl/test-pdf.xpl:2 unbound Error : Pipeline failed: err:XE0001: Output port 'result' on anonymous step declare-step at file:/C :/Sesi/personal/xml/xproc/calabash-0.9.14/calabash-0.9.14/xpl/test-pdf.xpl:2 unbound Unknown error

Posted by Sesi Bhushan on 31 Jul 2009 @ 12:22pm UTC #


I am not amused.

Why not clearly state everywhere that calabash is PHONING HOME messages to you everytime I use calabash?

Why not state everywhere that you log my IP at each use of calabash?

Why not clearly state how to turn this off?

Posted by nothere on 03 May 2010 @ 08:52am UTC #

I'm sorry you didn't find the documentation page where this is clearly explained:

It's a link straight off the documentation page and includes instructions for how to opt out.

In retrospect, it could be made more clear and I will attempt to do so. (You didn't leave any sort of contact information with your comment or I would have apologized more personally for any inconvenience.)

Posted by Norman Walsh on 04 May 2010 @ 10:43pm UTC #
Comments on this essay are closed. Thank you, spammers.