[oXygen-user] keep.relative.image.uris 0 being disrespected (???)

Chris Chiasson
Fri Aug 18 13:38:20 CDT 2006


I reinstalled Eclipse from the official page without WPT. It seems
oXygen retained its transformations (which makes sense given what
George Cristian Bina said about where transformation data is stored).
I made new ones from scratch just to be sure.

After further reading of the help, it seems that the parameters and
default listed in oXygen's transformation dialog are programatically
extracted from the referenced xsl stylesheet. I notice that it always
lists keep.relative.image.uris as 1, despite what my customization
layer says in common.xsl.

The transformation still results in broken image uris.

I have moved my stylesheets and auxiliary files into a subdirectory of
the project and placed the whole thing online (less than 100 KiB
uncompressed) at

http://chris.chiasson.name/temp/Engineering_Optimization_Documentation/

If a local copy would work better, I also made:

http://chris.chiasson.name/temp/Engineering_Optimization_Documentation.zip

I included a scenarios.properties file with all of my transformation
information.

The xml file being transformed is TestHarness.xml. The ./xml/xhtml.xsl
file is the one that serves as the entry to the customization layer
for unchunked xhtml output.

The two transformations I used to create TestHarness.xhtml and
TestHarness.html start with the name TestHarness.

Does any of this data indicate what I did wrong or what's going wrong?
How does the transformation turn out when you use xsltproc directly?

On 8/18/06, Sorin Ristache <> wrote:
> Hello Chris,
>
> It works for me with oXygen 7.2 and Saxon 6.5.5. If I use the default
> value 1 for keep.relative.image.uris the value of the graphic/@fileref
> attribute present in the DocBook source document is preserved as the
> value of img/@src in the XHTML result. If I set keep.relative.image.uris
>   to 0 and one of the ancestors of graphic/@fileref has an xml:base
> attribute then the value of the xml:base attribute of the topmost
> ancestor with this attribute is used as base URL to resolve the
> graphic/@fileref value. The resolved path is placed in img/@src in the
> XHTML result as with the default value of keep.relative.image.uris. If
> there is no ancestor with an xml:base attribute then the
> graphic/@fileref value is preserved. This algorithm is executed by the
> template called "relative-uri" in
> [oXygen-frameworks-folder]/docbook/xsl/common/common.xsl. This is true
> for both the oXygen standalone distribution and the Eclipse plugin one.
>
>
> Regards,
> Sorin
>
>
> Chris Chiasson wrote:
> > Hi, I have a stylesheet customization layer that sets the
> > keep.relative.image.uris parameter to zero. (therfore, they are
> > resolved against xml:base)
> >
> > I verified that the customization layer works by using xsltproc (I
> > have it on the Samba server that hosts my Windows files) on the xml
> > source files and xsl customization layer (and yes, the xml catalog on
> > the server points to docbook-xsl 1.70.1 as the current release).
> >
> > xsltproc --xinclude -o Engineering_Optimization.xhtml
> > ~/.xml/alpha/xhtml.xsl Engineering_Optimization.xml
> >
> > When I transform with oXygen 7.2 and saxon 6.5.5 using the same
> > customization layer (by placing file:/Z:/.xml/alpha/xhtml.xsl in the
> > XSL url field), the file comes out ok, with the exception that the
> > uris have not taken xml:base into account, so they don't point to the
> > correct folder (and thus the images don't show in Firefox).
> >
> > I have tried setting the parameter keep.relative.image.uris to 0
> > inside oXygen, which doesn't work either.
> >
> > I created my transformation by cloning the DocBook HTML one and
> > editing it to point to my xsl layer along with outputting to
> > ${cfn}.xhtml.
> >
> > I am going to go download the regular Eclipse distribution instead of
> > the Eclipse + Web Platform Tools distribution. If that fixes it, then
> > I'll post to this thread again.
> >
> _______________________________________________
> oXygen-user mailing list
> 
> http://www.oxygenxml.com/mailman/listinfo/oxygen-user
>


-- 
http://chris.chiasson.name/



More information about the oXygen-user mailing list