Re: jrb's help ideas

[Chris Lyttle <chris wilddev net>]

On 3 Sep 2001, Mikael Hallendal wrote:

> > The whole `help browser' concept is flawed. Why can't the help be made
> > available via http so any web browser can be used? Hell, you could
> > write it in perl or python if you wanted.
>
> I do not agree with you hear, while providing things for use in any
> browser (static HTML-pages) is good in some way (which you mention
> here). It's bad in others, like for example, you can make a much better
> and easier to use TOC if you for example use a Tree-widget.

Its my turn to disagree with you here, what David is talking about doesn't
have to be static, hell it doesn't even have to converted from XML/SGML.
In the web server world dynamicly generated websites are preferred over
static pages. There is no reason why a help system couldn't talk to a
'help' webserver and use all the scrollkeeper, TOC, glossary, etc features
we can think up. As David suggests this would be so much more powerful as
it would be accessible from the command line via a browser such as Lynx
and in X.


> > We need a system level server that provides access to a help database
> > that is not so gnome or kde specific. It should be able to serve any
> > document in the scrollkeeper database transparently to any browser can
> > supports html and http post/get.
>
> We have actually started to look at a solution like this at work
> (www.codefactory.se) and will start look into this issue next week.
> The ideas sprang out from DevHelp, and having a middlestation between
> the actual documentation and the viewer. So that you can get information
> from man/html/sgml/info/whatever and output them in
> text/html/xml/whatever...

This sounds like almost exactly what David is talking about. Programs like
Devhelp, Encompass, Nautilus or whatever should be able to access this
server as well as a generic html browser.

Chris


-- 
"When I see an adult on a bicycle, I have a hope for the human race."
   -- H.G. Wells