An anomaly in DocBook profiling came up on the Oxygen users mailing list, but it is actually a DocBook XSL issue, so I'm posting the problem and solution to the docbook-apps mailing list as well. From: "Sorin Ristache" <sorin@sync.ro> To: <oXygen-user@oxygenxml.com>
We did not use profile-docbook.xsl because it does not apply all the profiling attribute values always correctly for complex DocBook documents. We tested that version of profiling stylesheet on a complex document and it allowed in the output some subsections which did not have the profiling attribute value set as the profiling parameter of the transformation. The profile.xsl stylesheet did not have this problem so we used that in the Oxygen scenarios.
With Sorin's help, I tracked down the source of this problem, because it did not seem right to me that the profile-docbook.xsl stylesheet would produce a different profile from the profile.xsl stylesheet. It turns out to be an issue with customization of the DocBook HTML chunking stylesheet while profiling, and does not seem to be documented anywhere, so I'm posting the solution here for the record. Because DocBook's chunking stylesheet relies on XSL import precedence to separate the chunking functions from the formatting functions, a customization of the chunking stylesheet requires a particular setup that imports and includes various DocBook XSL modules in a certain order. That process is described here: http://www.sagehill.net/docbookxsl/ChunkingCustomization.html To customize the chunking stylesheet for single-pass profiling, my book says to import "profile-docbook.xsl" instead of "docbook.xsl". But that is not sufficient for chunked output. It is also necessary to replace the reference to "chunk-code.xsl" with "profile-chunk-code.xsl" in the customization file. The difference is that chunk-code.xsl matches on the original document nodeset, while profile-chunk-code.xsl matches on the profiled node set. If this substitution is not made, then the profiling won't work correctly for customized chunked output. I'll add this information to the next edition of my book. Bob Stayton Sagehill Enterprises bobs@sagehill.net ----- Original Message ----- From: "Sorin Ristache" <sorin@sync.ro> To: <oXygen-user@oxygenxml.com> Sent: Monday, June 06, 2011 6:39 AM Subject: Re: [oXygen-user] R: Additional XSLTstylesheet in docbooktransformation
Hello,
We did not use profile-docbook.xsl because it does not apply all the profiling attribute values always correctly for complex DocBook documents. We tested that version of profiling stylesheet on a complex document and it allowed in the output some subsections which did not have the profiling attribute value set as the profiling parameter of the transformation. The profile.xsl stylesheet did not have this problem so we used that in the Oxygen scenarios.
Regards, Sorin
Jirka Kosek wrote:
On 6.6.2011 12:03, Adrian Buza wrote:
Oxygen 12.2 added out-of-the-box support for DocBook profiling(conditional text) in the default transformation scenarios. This means that in the default transformation scenarios two stylesheets are applied to the original XML file: a generic profiling stylesheet(${frameworks}/docbook/xsl/profiling/profile.xsl) and the specific transformation stylesheet(e.g. PDF: ${frameworks}/docbook/xsl/fo/docbook_custom.xsl).
I'm just curious -- is there any reason why you haven't used profile-docbook.xsl instead, which does profiling and actual transformation in a single step?
Jirka
oXygen-user mailing list oXygen-user@oxygenxml.com http://www.oxygenxml.com/mailman/listinfo/oxygen-user