DOCBOOK-APPS: writing a condensed form of docbook

["Robert P. J. Day" <rpjday@mindspring.com>]

  while i know there are numerous XML editors to make one's
life easier, one of my issues is that, in the end, the 
docbook that is produced is virtually impossible to scan
(although i'm sure numerous people will tell me that that's
not the point, you're not *supposed* to be able to scan it,
just its transformed output.)

  i like to write my documents and manuals in condensed
form, and it would be nice to quickly to go from that to 
final docbook.

  to that end, one of my techniques was to find 2-letter
abbreviations for almost all of the docbook elements i 
used, and create a corresponding emacs abbreviations
file.  

  that made typing easier, but still left me with unreadable
docbook, so i'm considering writing my stuff in what i'll
call "pidgin" docbook, and having a post-processor (sed
or perl) turn it into real docbook.

  as an example, instead of writing

  <itemizedlist>
   <listitem>
    <para>
     blah blah
    </para>
   </listitem>
   <listitem>
    <para>
     woof woof
    </para>
   </listitem>
  </itemizedlist>

i would just write

<il>
<li>
blah blah
</li>
<li>
woof woof
</li>
</il>

or possibly condense it even further, who knows?  it would be nice
to be able to write my docbook sort of "stream of consciousness",
throwing out lists and so on, knowing that, when i was done, i could
just run it all thru my post-processor to get the real thing.

  i'm sure several people here are gagging at this point, but i
found that using the emacs abbreviations worked pretty well since,
given the subset of docbook elements i was using frequently, i could
assign an intuitive and unique 2-letter abbreviation to almost all
of them, like

il = itemizedlist
ol = orderedlist
li = listitem
em = emphasis
vn = varname
fn = filename
pl = programlisting
sc = screen

and so on.

  is there anyone who's tried this?  and maybe even wants to 
share their post-processor or something i can hack to get this?
it certainly can't be hard -- perl would make short work of this,
don't you think?  

  comments?

rday