Re: Gnome Help Browser

[Chris Lyttle <chris wilddev net>]

On 19 Aug 2001 23:47:58 -0400, David Merrill wrote:
> On Sun, Aug 19, 2001 at 06:25:47PM -0600, John Fleck wrote:
> > On Sun, Aug 19, 2001 at 03:31:07PM -0400, David Merrill wrote:
> > > > 
> > > > I think Sasha is right here. A help browser needs to understand what
> > > > do with a "gnome-help:foo", "man:foo" or "info:food" URL, or linking
> > > > between and among documents will not work. You're right that if we
> > > > do gnome-db2htl3 right it will output valid html, but the browser
> > > > still needs the additional ability to know that when it sees a GNOME
> > > > help-related url it needs to hand it off to the help system.
> > > 
> > > I don't see how that follows. Yes, the *system* has to know how to
> > > translate man:foo, but that doesn't mean the browser has to do it. Why
> > > can't the browser talk to the library using http post?
> > > 
> > 
> > David, could you explain in more detail what you mean here? I'd love a
> > system that could use any browser rather than one that has to be
> > specially designed to handle our various help urls.
> > 
> > How would an arbitrary browser that did not know anything about
> > "gnome-help:foo" or "man:foo" or "info:foo" urls use http post to
> > access the help system when it saw such a url? Or, alternatively, is
> > there a different way we could construct our urls so an arbitrary
> > browser would know, "Hey, this isn't an ordinary http or file url, I'd
> > better send it off to the help system."
> 
> The scrollkeeper system is doing the job of abstracting the
> documentation into a library. Why not build a layer on top of that
> that serves the help via http, and then any browser can use it,
> including a remote browser. It performs format conversions, searching
> capabilities, etc.
> 
> Imagine a company or project (even the GNOME project or the LDP)
> serving custom help repositories to which you can subscribe. Remember
> that scrollkeeper will support remote documents in a future release.
> Mny excellent sources on the net could be indexed and merged
> seamlessly into your help environment.
> 
In my mind this is a much better approach. Many projects now include 
a mini 'webserver' or such on a particular port number. It would just be necessary
to code a GNOME-help webserver that uses a certain port for
communication and tell the browser to query that port for the help
pages. That way you would only need to worry about having the webserver
translate the docs into html for the browser and interface with
scrollkeeper. This would also be useful in that the webserver could
'redirect' when necessary to access remote information. All this is
standard in the webserver world, all we would need to do is make the
interface to the help docs.

Chris