>
> Apparently there's this Web publishing tool called Trellix. I'm not
> really tempted to use it, but I do think there are some things that
> could be learned from it, so I thought I'd mention it and give a
> pointer to a doc prepared with it.
>
> Anyway, I don't know much about it, but the founder of the company
> that develops it took the horribly organized Starr report and re-did
> it with Trellix, and the result is pretty outstanding.
Yes, it is a wonderful job.
I tried to look at the version that Netscape had made avaiable,
but gave up due to the delays in loading new pages.
All that surrounding guff of advertising via animated images
made it an impossible read.
> Here's a short column that talks about the Trellix guy re-formatting
> the document:
>
> http://www.zdnet.com/pcweek/stories/columns/0,4351,351939,00.html
All the principles mentioned are available with LaTeX2HTML,
except that:
1. frame support needs to be refined, and updated for HTML 4.0
2. methods need to be developed to put the navigation panels
into frames; this has been discussed previously, but the
work hasn't been done.
3. with LaTeX source, the headings and footnotes/cross-references
are already marked. For a document where this is not so,
there needs to be some pre-processing to locate these
and insert appropriate markers (e.g. LaTeX macro-names)
before doing the main processing.
4. create appropriate images for the new navigation techniques.
Point 3 is essentially writing partial translators:
Word, Ventura, Ami-pro, etc. --> LaTeX
I say *partial*, because you don't need much in the preamble, as the body
won't be using macros from existing packages. But LaTeX2HTML doesn't
require the source to be a complete LaTeX document.
For a specific job, some ad-hoc perl scripts, written specifically for
that job can be extremely effective.
Such ad hoc techniques can be used also to *fix up* the (supposed) HTML
code produced by the commercial packages. Frequently these, especially
older versions, produce invalid code with badly nested tags. Sometimes
this is due to the software, or it can be due to poor use of it.
In any case, such errors can be detected are fixed, using Perl's
pattern-matching abilities.
> And here's the re-done Starr report:
>
> http://www.trellix.com/icreport/
>
> There are pointers in there to the original report, or you can find it
> all over the web.
>
> Like I said, I'm not really tempted to use Trellix, but I do think
> there are some things that could be learned from it, in particular the
> way it treats footnotes and that little "overview" map at the top.
Martin Wilck's 'frames.perl' module already implements this idea for footnotes.
Here is the 1996 version of the LaTeX2HTML manual, processed
using frames:
http://www-math.mpce.mq.edu.au/~ross/latex/netscape/manual/
For a chapter with footnotes, have a look at:
http://www-math.mpce.mq.edu.au/~ross/latex/netscape/manual/Snode1.html
> Thoughts?
>
Get a student to update this work.
Copy ideas seen on other Web pages; write appropriate LaTeX code to
insert them into LaTeX2HTML-produced pages...
...and include HTML comments attributing the original source.
Hope this helps,
Ross Moore
