Re: Idea : common dir and tree (Rep:Re: Why not create packages?)

["Greg Ferguson" <gferg@hoop.timonium.sgi.com>]

On Jul 18,  9:12pm, <guylhemlistes@ifrance.com> wrote:
> Subject: Idea : common dir and tree (Rep:Re: Why not create packages?)
>
> [ plain text
>   Encoded with "quoted-printable" ] :
> Perhaps LDP should specify a distribution-neutral
> > location for all the HOWTOs, such as /opt/doc
>
> We have been talking about this with KDP and GDP.
>
> Basically, for a unified documentation browser/search
> engine, we need a standard.
>
> My idea would be putting the following hierarchy in a
> dir, for ex /opt/doc :
>
> lang/jp/html
> lang/jp/pdf
> lang/de (etc)
> html/
> pdf/
> pdb/
> txt/
> info/
> man/
> (...)
>
> We could just provide link for dirs (for ex,
> /usr/man/man1 to man/man1) or for files (for ex
> /usr/doc/XXXX in txt/XXXX) or put plain files in that
> dir, as you suggested.
>
> In both cases, /opt/doc would look the same on each
> distribution, wherever the documents are really put
> in, and a documentation browser or a hand browsing
> would be very easy, all the documents being available
> in a single dir..
>
> Likewise packages would only have to add their
> documentation there to be automatically added to the
> browser "index" and "search base", regardless where
> it is installed.
>
> For ex, if staroffice did provide pdf documentation
> in /usr/doc/some/lost/directory/somewhere/doc.pdf, it
> would be as simple as providing a link in
> /opt/doc/pdf/staroffice to add its documentation to
> the database.

I'll throw out what we came up with at SGI. I wrote a short
paper on this topic (SGI's linux documentation delivery), and
*tried* to maintain and use whatever standards were available.

Granted, these conventions are targetted more toward "manuals",
but nonetheless, I/we tried to address some of the issues
Guylhem brings up (such as resource discovery).

-------

Books are installed in:

/usr/doc/sgi/<prod>-<vrs>/<book_shorttitle>-<rev>/{html,ps,pdf,...}

as in:

/usr/doc/sgi/sgi1400-1.0/1400Servr_UG-001/{html,ps,pdf,...}/

Within each specific book instance directory there is an XML-based,
metadata-rich, Document Description File (DDF). That file contains,
among other things, information about each of the various available
formats for the content piece (HTML, PDF, Postscript, etc.).

Ideally a search/browse engine would crawl, examine, and collect
up the various DDF files that are available and use those
to trigger indexing and catalogging of data, etc. -- very similar
to the concepts used in the Dublin Core, metalab's OMF, etc.

No sym-links, nothing of that nature required. The difficulty
lies in knowing where to "find" these resource-description files.

-------

regards,

-- 
Greg Ferguson     - s/w engr / mtlhd         | gferg@sgi.com
SGI Tech Pubs     - http://techpubs.sgi.com  | 
Linux Doc Project - http://www.linuxdoc.org  |


--  
To UNSUBSCRIBE, email to ldp-discuss-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org