Re: entities in Nautilus (was Re: reduce redundancy?)

[<jfleck inkstain net>]

On 22 Aug 2000, Karl Eichwalder wrote:

> John Fleck <jfleck@inkstain.net> writes:
> 
> > Using sx or sgmlnorm might work, but it would involve essentially
> > keeping two sets of docs - the writer's original, using entities, and
> > the final distribution version, which has been normalized to remove
> > the entities. This seems like a huge headache.
> 
> The writer delivers SGML; this is the distribution version (the tar.gz
> file).  `make help' will produce installable SGML files and `make
> install-help' will install these files.  `make html' will produce HTML
> and `make install-html' will install HTML browsable with regular HTML
> browsers.  `make print': PostScript (via DVI?); `make pdf', `make
> braille', etc.
> 

Sorry if I'm being dense, but it seems like 

We have:

original.sgml, which is what the writer writes (plus some other files that
are included as entities, for example)

This lives in cvs.

So when the package maintainer is actually building a package, they run
'make help' to create installable.sgml (the normalized, entity-free
version) as part of the package? "original.sgml" then remains in cvs, and
"installable.sgml" goes out unto the world, that Nautilus may read it.

Or Karl, are you suggesting that `original.sgml' goes out in the
distribution package and the `make help' step happens on the user
end? That would require all the users to have sx or sgmlnorm or something,
which I think is a bad idea.

Cheers,

John