[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Our Discussion on Fedora-docs [Fwd: Re: Fedora Documentation Search Engine]

Dear all,

I thought that maybe you guys would have some input on this?


---------------------------- Original Message ----------------------------
Subject: Re: Fedora Documentation Search Engine
From:    "Stuart Ellis" <s ellis fastmail co uk>
Date:    Fri, January 28, 2005 3:24 pm
To:      "For participants of the docs project"
<fedora-docs-list redhat com>

On Fri, 28 Jan 2005 14:39:32 -0000 (GMT), "Gavin Henry"
<ghenry suretecsystems com> said:
> <quote who="Stuart Ellis">
> >
> > On Fri, 28 Jan 2005 12:56:58 -0000 (GMT), "Gavin Henry"
> > <ghenry suretecsystems com> said:
> >> Dear all,
> >>
> >> Has there been any discussion about this?
> >>
> >> I thinking along the lines of htdig/swish-e that indexes all man
pages/howto/README after every rpm is installed.
> >>
> >> Something like a post entry in the spec file, or similar.
> >
> > I haven't seen any on this list, but essentially this is a function of
having a desktop search/indexing engine since there isn't a common
format to this stuff.  The next release of yelp (GNOME help browser)
will display info and man pages, but can't index the random txt, html,
pdf etc. that goes into /usr/share/doc/.
> We heavily use Swish-e (www.swish-e.org) for the fileservers we install.
This can handle html, xhtml, txt, pdf and the like.

I've just looked at the page now, but it looks like a very useful bit of

> Maybe something like this or a updatedocdb crontab, like slocate has and
we just put the standard doc pathnames in a config file. The customize
> web search page.

My understanding is that this is where things become quite technical and
problematic. There is always a disjunct between what's actually on the
disk and what is listed in the index.  On portable computers it's harder
to guarantee that batch indexing jobs will be completed regularly as well.
 Beagle and Spotlight use kernel hooks to update the index as changes
occur.  I suppose that RPM packages could run post-install scripts that
update a document registry (kind of like you suggested) would be a
relatively non-invasive way to implement something.

I definitely think that the default start page of the Web browser could be
used very effectively by the docs project.  The details of
implementing an index might be a good question for fedora-devel.


Stuart Ellis
s ellis fastmail co uk

fedora-docs-list mailing list
fedora-docs-list redhat com
To unsubscribe:

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]