completeness check fails on xrefs

[BrianJ]

All,

We have a large document that we assemble from several smaller docs
managed and versioned independently. The large doc has a cross
reference where the ID is in one of the smaller docs and the IDREF is in
another. The problem is, our CMS checkin process (which we coded
ourselves), runs a completeness check on the doc first, and only
proceeds with the checkin if the check succeeds. The smaller docs with
the ID and the IDREF obviously fail the completeness check since the doc
is missing either the ID or the IDREF. Is there a way to get around
this while still running the completeness check?

Thanks,
--



Brian Jensen
bjensen@bluelid.com

[bibach]

If an attribute of type IDREF contains a value that is not present in an
attribute of type ID in the same document, that document is not valid XML.
I don't think there's much of anything you can do to get around that, and
I'd be very surprised if Editor provides an option to skip that check.

You could, however, modify your DTD to declare the IDREF attribute to be
CDATA, instead.

-Brandon 🙂

[BrianJ]

Thanks for your reply Brandon. As I understand it (I'm asking for a
coworker), the ID/IDREF types are required to get cross reference
functionality in Arbortext. Am I mistaken? Anybody out there with
external cross references that could explain how they are doing it?

Thanks,



Brian Jensen
bjensen@bluelid.com

[ray.anderson]

We have a similar situation where we have a larger doc made up of
smaller docs. We wrote a tool to do a ID/IDREF check of the larger doc
and we "turn-off" the ID/IDREF check of the smaller docs because we know
that they may not "pass" because of the references to "other" docs.



Could you do something along those lines?



Ray

[BrianJ]

Thanks Ray. Is there a way to turn off only the ID/IDREF check when
doing the completeness check, or do you just not do the completeness
check on the smaller docs?




Brian Jensen
bjensen@bluelid.com

[ray.anderson]

In our case, we do the ID/IDREF check with an OmniMark program. When we
do the parse of the smaller docs we use the "with id-checking false".
Our tool that does the ID/IDREF checking of the larger doc also uses an
OmniMark program.



You had stated that you had coded the check-in process so I thought you
may be in control of the ID/IDREF checking.



Ray

[bibach]

Generally speaking, the only value you get from IDREF attributes is error
messages from your parser if the referenced ID doesn't exist. Any special
treatment, like linking or cross-referencing behavior, in either the
authoring or publishing environments, is due to specific code in those
tools, and that code is just matching up strings in the source and target,
so it doesn't usually care about IDREF vs. CDATA. The cross-reference
elements in DITA, for instance, use CDATA attributes and support cross-topic
references. However, Arbortext has significant custom code to support
these.

My guess is that changing your IDREF attributes to CDATA probably will
require few, if any, modifications to your publishing tools. As for the
behavior within Editor, such as displaying generated text derived from the
cross-reference target, unless you have some custom code in place for that
now, you probably aren't getting it, anyway.

It would probably be worth doing a test with the attributes as CDATA and see
what functionality breaks, if any. You can always come back to the group
with the specifics for help.

-Brandon 🙂

[BrianJ]

Ray and Brandon,

Thanks again for your help. Our CMS is custom-made by us, but we run
Arbortext's completeness check before we let the doc get any further in
the checkin process. We are using Styler to manage the style of the
Editor view and composed output, and PE to manage the composition. My
coworker said he did try getting the cross references working without
the ID/IDREF types, but it didn't work. With the ID/IDREF attributes in
place he does have the cross references working properly, including the
gentext.
Ray, just to confirm, your "with id-checking false" parsing is referring
to OmniMark, right?

Thanks,



Brian Jensen
bjensen@bluelid.com

[ray.anderson]

Yes, the "with id-checking false" does refer to OmniMark.

[bibach]

Hi, Brian...

Okay, so let me see if I understand. You author a set of document
components, working on just a single component at a time inside Arbortext
Editor. These components sometimes have unresolved IDREF values because the
target ID is in another component.

When it's time to publish, some set of components gets assembled into a
complete document, which conforms to the same DTD as the individual
components, and a Styler style sheet is applied to it. The complete
document has no unresolved IDREF values because all of the components are
merged into a single file. Is all of that correct?

Styler seems only want to let you use an IDREF attribute for specifying the
target for a link or cross-reference, yet some of the distributed style
sheets (such as for DITA) use CDATA attributes. If your Styler sheet is
handling your cross-references correctly and the only issue is the errors
you get about missing ID values, try changing the attribute on your
cross-reference element to CDATA and see if everything works the same, minus
the "missing ID" errors.

If this doesn't get you where you need to go, we'd need some more
information about your setup (markup examples, DTD details, etc.) and how
exactly it "didn't work".

-Brandon 🙂