Hello Martin,
I've been lurking on the discuss list and have fixed the text following some
suggestions I received.
Please find attached the final version of my proposed Howto; I hope it's going
to be accepted.
All the best,
Guido =8-)
This version should not be present here.....if the LFS folk wish to contribute
it, then, it should go right on top of the older version. I did not want to
make that choice for them, but having the duplicate here is also not good, so
I'm removing it.
added a DocBook 5.0 declaration and, after failing validation with jing,
changed the anchor location to be inside the <firstname/> element.
Validates now!
(and restoring commit 78ec6080e6 which
suppresses the dbhtml PI for document naming)
Assembly-HOWTO.xml was written as a DocBook 5.0 document, however it sported a
declaration of a DocBook XML 4.5 document. This confused me and I changed all
of the <xlink:href/> elements to <ulink/> elements.
reverts commit 13943cb7fb
It is not valid to have <indexterm/> elements directly as a child of a
<varlistentry/>, so I have moved them inside the <listitem/>. I tried putting
them inbetween <varlistentry/> elements (no good).
It is not valid to have <indexterm/> elements directly as a child of a
<varlistentry/>, so I have moved them inside the <listitem/>. I tried putting
them inbetween <varlistentry/> elements (no good).
It is not valid to have <indexterm/> elements directly as a child of a
<varlistentry/>, so I have moved them inside the <listitem/>. I tried putting
them inbetween <varlistentry/> elements (no good).
there were two locations where CDATA was outside of an enclosing </para> or
inside another element that could not contain it (according to the DocBook XML
element nesting model).
an <indexterm/> cannot be a direct child of an <abstract/>, so I moved it
inside of the <para/>
neither can <command/> be the direct children of a <sect1/>, so I moved them,
each inside a <para/>
An <abstract/> element can contain a <para/>, which can contain an
<itemizedlist/>, but the <itemizedlist/> may not be a direct child of an
<abstract/>.
there was a mismatch between the root element and the doctype declaration;
assuming that the root element is correct, synchronizing these two changes
fixed the validation problems
the doctype declaration had conflicting entries; the public identifier said
4.1.2 and the system identifier said 4.0; I would not have noticed this,
except that xmllint was complaining about the <author/> element containing an
<email/> (invalid!)
this was easily resolved by changing the version of this document to be a
DocBook XML 4.2 document
DocBook 4.x appears to be fairly picky about order in <author/> elements and
the document will not validate without the order being correct. Fixed.
Also, in the <revhistory/>, there were plenty of places where multiple
<revremark/> elements needed to be wrapped in a <revdescription/> along with a
bunch of <remark/> elements instead.
first, this is not XSL and in DocBook 4.x, these extra namespaces confuse the
processing toolchain (may be different in DocBook 5.x); next, xmllint refuses
to validate the document because of the order of elements in, for example the
<author/> elements and the <affiliation/> elements; just flipping these in
most cases did the trick
added a bunch of corrections for <itemizedlist/> <listitem/> and <para/>
nesting to match the DocBook content model more strictly; this allows full
validation of the document