Matthew P. Barnson wrote:

> On Friday 05 January 2001 16:14, James Green scratched this in the dirt:
> 
> 
>> How do people using only Netscape/Mozilla Composer to produce their
>> pages, and get these saved html pages into DocBook format without
>> needing any further tools or knowledge or in fact effort?

>   Heh Heh, that's the kicker.  The way Linuxdoc.org has gotten around that is 
> to ask people to submit documentation in whatever medium is convenient, then 
> have a staff of volunteers convert the file into DocBook.  It's worked so 
> far, for hundreds of HOWTO and Guide documents.

Do our "target authors" (mozilla.org|netscape.com staff, plus 
contributors) match Linux documentors in ability to use the DocBook 
tools and have the time to learn it, and install what is needed?

[ snip ]

>   The easiest way to do DocBook is to have an editor which supports SGML or 
> XML.  There are some commercial plugins for MS word which support it, Adobe 
> Framemaker supports it...

Niether of those reads "Netscape|Mozilla Composer" for any version. I 
somehow doubt that those staff on Win32 are going to migrate to using 
Emacs (I can't use the damned interface and I'm used to vim!), but they 
_may_ use framemaker or Word (although that would mean them having the 
software and not mind using it).

> I personally use PSGML mode in Emacs to write my 
> docs, but that's a pretty geeky way to do it (kinda' like writing HTML in a 
> validating text editor).

Well I'd prefer not to use emacs full stop :). But I also wouldn't want 
to spend 3 hours looking for the incorrect tag :).

>   The key issue there, though, is people have gotten used to the idea of 
> "What you see is what you get" rather than "what you see is what you mean".  

I seriously doubt if we can train the staff to work out what they mean 
in DTD context.

> Once you've passed that stumbling block, and get people to understand that 
> they need to define the type of text they are putting in, not how it's 
> supposed to look, the rest is cake.

Have you seen the number of possible types/contexts of the information 
on the current site? I'm pretty sure that we'd need to more strictly 
control context and layout using DocBook than are now. For example, we 
would lose text colours in some parts, gain it in others.

>   Many HTML people understand this instinctively, because HTML can be defined 
> as simply an SGML DTD...

Indeed. But how many of the mozilla/netscape staff could actually write 
their web page using a simple text editor? How many rely on Composer for 
the job, or whom would simply refuse to use anything other than a 
graphical page builder? I may be doing them an injustice, I have no way 
of telling since I've never met them, but going by experience of people 
being shocked that I actually refuse to use graphical web authoring tools...

>   My suggestion: Get a group of volunteers who will do DocBook conversions.

 From what type of source? Complete webpage taken from mozilla.org and 
edited? how do we know what's changed? Sure we could use diff, but then 
with graphical tools changing other bits of source automatically 
("cleaning it up"), we could never be sure.

Can we guarentee that a change they make can indeed be put into our 
docbook format? If it's a fancy form hooked up to something, can we put 
that in a docbook source?

> Anybody who understands HTML can easily understand DocBook.  It's pretty 
> trivial to convert most documentation, but the decisions on how to convert 
> need to be made by a human, not a machine, because tagging semantics differ 
> according to author preference.

I'm worried that we may spend a long time in getting a DTD to give us 
the flexibility needed by the authors, and may be this isn't worth the 
effort. Remember, the HOWTOs are very set in there layouts with very 
little markup, and hardly any colouring, would a DID for the mozilla.org 
site prove a huge task?

>   I volunteer to be the first volunteer.  Get at least two more people and a 
> submission mailing list, and you are in business.

If you care to take your news reader back in time, around late 1998 I 
see a large thread discussion exactly what we are here. By the looks of 
the mozilla.org site in cvs, i gather they settled on not using DocBook. 
The posts are still on the server, in this group.

James Green


Reply via email to