This is the mail archive of the
docbook-tools-discuss@sourceware.cygnus.com
mailing list for the docbook-tools project.
Re: [Fwd: Evolution of the DocBook tools]
* Eric Bischoff <ebisch@cybercable.tm.fr> wrote:
> Jochem Huhmann wrote:
>
> > There is another possible approach: A defined interface instead of a
> > defined directory layout.
>
> We could do both.
We even should. ;-)
> I have been reading your paper on the topic and it's a very
> interesting approach. But now we have an opportunity to
> define common directory names and catalog file names we
> should take profit of it.
We can define them (and we should), but we shouldn't rely on them. Look
at FreeBSD, at Solaris - we can make this as configurable as possible.
> The scripts I've been rewriting also try to guess the
> locations (a third approach), so they don't even need this
> directory layout either. But I still think a common
> directory layout would simplify things.
Guessing is guessing. What if someone has installed different
stylesheets etc. installed? I've some from sgmltools (which came with
jade and stylesheets and installs in /usr/local) and someone might have
different versions installed by different packages in different but
common locations. I don't like guessing.
> You're perfectly right. But as I said before, we can fight
> on both fronts.
Yepp. Let's try to find the one and best locations for all files and
directories, define this as default and standard and get it sacrified by
the LSB - and make it configurable for the poor souls with systems not
compliant to this standard. I would like to make no assumptions to rely
on wrt file locations. This is also a good test for being a valuable
tool for SGML-environments: if we make the script complete and
self-containing, you can query it anytime for the real, "official"
SGML-environment on the system it runs on. And use it to manipulate this
environment without having to know if it is compliant to the standard or
not. If "db --install-catalog /mnt/cdrom/fancy.cat" just adds these
entries to the catalog, you don't need to know where the catalog is at
all. And if "db --query catalog" returns the location of the catalog,
this is a great thing for other software that needs to know where the
catalog is (PSGML anyone?). It can just ask.
> Attached the new scripts (with the backends system) with the
> corresponding man page drafts.
>
> If a charity person can convert these drafts into good
> docbook and send me the result, he would make me gain some
> precious time. Thanks in advance.
I will look at it later. Thanks for the effort anyway,
Jochem
--
Hi! I'm a .signature virus! Copy me into your ~/.signature to help me spread!