Add a Description of Dublin Core Outputs for XHTML Based on DITA XML Inputs

[KeithIXIASOFT]

When someone outputs default XHTML from DITA content, DITA metadata is converted to its Dublin Core equivalents. This information is not currently covered in the DITA-OT documentation.

After doing some research as to how this works, I have determined which DITA tags produce their equivalent Dublin Core outputs in XHTML. The attached PowerPoint slides
DITA-OT and Dublin Core.pptx describes the relationship between DITA and the Dublin Core it outputs that I have discovered.

I'd recommend the following topic breakdown to cover things off:

• What is Dublin Core?
• Dublin Core Expression in XHTML Output (from the DITA-OT)
• DITA Metadata and Dublin Core Expressed by the DITA-OT

I would be happy to help, and I would also consider writing draft topics on this subject to contribute to the DITA-OT docs.

[robander]

There's one note of caution I'd have here. I don't think we have any other spot in the documentation where we explicitly document the mapping of a DITA element to a specific HTML style or element. There's good reason for that - the renderings are not absolute (not part of the spec), commonly overridden, and might change from release to release.

Having our documentation contain a list of specific mappings seems to provide a stronger-than-warranted guarantee that the mappings are A) correct, and B) set in stone.

[KeithIXIASOFT]

I take your point. Perhaps a more generalized addition that talks about what Dublin Core is and describing only the outputs that are available without explicitly tying them back to specific DITA tags?

[robander]

One possible approach - we could update the X/HTML topics to state that for these transform types, DITA metadata elements are currently rendered as Dublin Core metadata. If we need a whole topic devoted to it, we could instead make a "DITA Metadata" topic under the "DITA specification support" section.

Either way - I don't think we should go in depth describing Dublin Core itself, but should instead link to dublincore.org for reference (maybe to http://dublincore.org/metadata-basics/ or some other high level overview page). After all, we're not experts on that initiative, and shouldn't try to describe a subject that will always be better described on their own site.

So - to grossly paraphrase, something like "When possible we convert DITA metadata to Dublin Core. It's a global initiative for encoding metadata, learn about it [here]. We currently do all of this for all of our HTML and XHTML based transform types."

[KeithIXIASOFT]

That seems like a sensible approach. I think that a statement along those lines definitely needs to be in the DITA-OT docs somewhere.

I will likely come up with a separate blog post that describes things in more detail from my research, if that's okay with you. It would be, of course, unofficial.

[robander]

I have no problem with a blog post on the subject. It shouldn't actually require that much research - the mappings aren't intentionally hidden, and should all be found in
plugins/org.dita.xhtml/xsl/xslhtml/get-meta.xsl

You can search for "DC." to locate any spot in there where we generate Dublin Core values. For most of those - probably not all, but definitely most - it will be clear from the XSLT context how the mapping is created. For example, one of the first you'll find in the file is this:

<xsl:template match="*[contains(@class,' topic/source ')]/@href" mode="gen-metadata">
  <meta name="DC.Source" content="{normalize-space(.)}"/>
  <xsl:value-of select="$newline"/>
</xsl:template>

This is generating the DC.Source metadata value, based off of *[contains(@class,' topic/source ')]/@href -- that is, based off of the @href attribute on the element with topic/source in the class value. In other words, <source href="this-maps-to-DC.source">.

[KeithIXIASOFT]

Very cool! I had not gone digging in the DITA-OT code, but now that you describe it is obvious that I should've looked there.

Is it worth mentioning where the DC output values are stored in the DITA-OT docs, or is that something that could also conceivably change over time/releases?

[robander]

I wouldn't mention that file in the docs - while it's possible we would even switch from Dublin Core to some other representation at some point, it's much more likely that mappings will move to a new file or that the file will be moved / renamed. Generally speaking, I'd expect any documentation of a specific XSL file name or location to get out of sync before long.

[KeithIXIASOFT]

That's a fair point. Given that I am hoping to bring you some optional Schema.org mapping of metadata values at some point in the future (fingers-crossed) this makes sense.

[stale]

This issue has been automatically marked as stale because it has not been updated recently. It will be closed soon if no further activity occurs. Thank you for your contributions.