Re: Alternate Proposals to Nautilus

[Gregory Leblanc <gleblanc linuxweasel com>]

On 11 Aug 2001 13:40:16 -0600, John Fleck wrote:
> On Sat, Aug 11, 2001 at 09:16:41AM -0500, Eric Baudais wrote:
> > 
> > The problem with making a new application is finding a hacker to code it.  
> > This is not an easy task.  We've had multiple hackers code the 
> > gnome-db2html[1-3] back end.
> 
> Drake nailed it here.
> 
> I have no problem with an alternative, lightweight browser that
> embodies the functionality we have now in Nautilus - a
> Scrollkeeper-generated list of docs, indexing-to-come. But if we want
> to go anywhere with this discussion, we need a hacker to build the
> thing. That's not easy, which is why I've ended up doing significant
> parts of gnome-db2html2 and 3 myself (and if anyone has seen my code,
> you understand what a bad idea that is :-)

Hmm, I guess I should make myself clearer in the future.  The reason
that I said that Nautilus was the help solution in GNOME 1.4 is that
g-h-b didn't get any hacking in that timeframe, while the Help stuff in
Nautilus got quite a bit.  Nautilus is far more feature-rich than the
old help browser.  If you turn off all of the fancy graphics, and do 1
or two other tweaks (like setting it to be very early (priority 10) in
the list of startup programs), nautilus is pretty bearable on my P-II
233 64MB machine.  It's not great, but certainly usable, and it gives me
a useful help sidebar (and that Uber-cool news sidebar).

OK, enough of that rant, onto productive things.  We need a way to view
help in GNOME 2.0 (duh!).  We have libxslt at our disposal, and moving
our docs to DocBook XML 4 will be a relatively painless task.  So, let's
assume that any HTML rendering program can be used to actually view the
docs.  We've also got Scrollkeeper, which gives us a categorized list of
docs, and a TOC that we can access without actually having to render the
document, and a whole bunch of meta-data that we can search on (such as
Author, Title, Abstract, etc.).  I've not done Scrollkeeper justice, but
that's the best I can manage at the moment.

Our help browser will need to leverage both of these technologies to be
effective.  Right now, the only possible help browser that can use
Scrollkeeper is Nautilus.  Most of the readers of this list aren't
hackers, so we'd probably have a hard time writing an application that
can take advantage of Scrollkeeper on our own.  We've also got to
maintain the code that converts from XML to HTML (libxslt does most of
this, but there's some other magic to be done as well).  We need a
hacker to maintain/write this code, and it's got to get done before the
end of the year.  Nautilus will need some changes to take advantage of
the enhancements to Scrollkeeper, but there are people familiar with
that code already, and a goodly portion of it is already written.  There
have also been people working on designing how the UI for Help in
Nautilus should look.

I don't think we'll be able to write something from scratch that can
beat what we'll get by improving Nautilus.  There's a lot of hacking
still going on with Nautilus, and ever release is better than the last.
Think about it.
	Greg, show me the code, Leblanc.