Plan of attack for XML changes

Matthew Burgess ca9mbu at hermes.sunderland.ac.uk
Mon Jul 28 14:21:40 PDT 2003


On Sat, 26 Jul 2003 16:46:44 +0100
Matthew Burgess <ca9mbu at hermes.sunderland.ac.uk> wrote:

<snip message to which no-one responded!>

Well, seeing as I received no objections to that plan I assumed it was
sensible.  I've gone ahead and made the changes outlined below.  The
updated sources are available at
http://osiris.sunderland.ac.uk/~ca9mbu/lfs-book/lfs-xincludes-WIP-20030728.tar.gz.

CHANGES
~~~~~~~

1. Any entities that were simply references to other files have been
   replaced with the equivalent XInclude instruction.
2. The division of content between files has been modified in places in
   order to allow each file to be a self contained valid XML document.
   Note that I've not actually validated them against the DTD yet as I'm
   constrained to a Windows environment that doesn't take kindly to the
   /usr/share/docbook/docbookx.dtd path.  Having said that, processing
   the book with `xmllint --noent --xinclude index.xml` doesn't show any
   well-formedness errors.

Note also that I've not touched chapter 6 yet, but it'll follow the same
format as the chapter 5 packages if noone sees anything wrong with the
current changes.  With the reduction in the use of entities it looks as
though the separate entities/package.ent files could be amalgamated into
one big file as all they contain is the version of the specific package.
 The other entities that were being used from those files were either
paths to other xml documents (and have hence been removed as per point
1. above) or were bits of data that weren't being re-used anywhere and
therefore didn't need to be entities.

Whilst I've tried to minimise breakage as much as possible the current
changes mean that chapter 4 and the appendix are broken due to their
reliance on the new entity management.

Hope someone can make use of this - I'll post a completed tarball in the
next couple of days with fixed chapters 4, 6 and appendix if there's no
complaints.

Best regards,

Matt.



More information about the lfs-dev mailing list