Subject: Re: Location of NetBSD DTDs
To: Hiroki Sato <>
From: Mike M. Volokhov <>
List: netbsd-docs
Date: 05/10/2005 12:54:37
Content-Type: text/plain; charset=US-ASCII
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Sat, 07 May 2005 03:29:29 +0900 (JST)
Hiroki Sato <> wrote:

> "Mike M. Volokhov" <> wrote
>   in <>:
> mi> But specified path to DTD is not available. Is this correct? Or it (a=
> mi> all descending paths) should be fixed to the:
> mi>=20
> mi> 	""
> mi>=20
> mi> Possible I've missing something?
>  I intentionally choose the different prefix from ones used in
>  the webpages in  This is because the URI in the
>  doctype declaration is used as a namespace and actually it
>  has nothing to do with the WWW server.  This URI is resolved by XML
>  catalogs in htdocs/share/xml/catalog-common.xml and
>  htdocs/share/xml/catalog.xml and then always points one of
>  local files.

Well, there are many software which wish to have DTD available
somewhere, and catalog.xml is not available anytime (for example, when
opening documents via Internet). For example, this might be online
markup validator (please try

>  This resolution provides some sort of flexibility and consistency
>  which is especially useful for localization.  For example,
>  "/lang/" in the URI can be replaced with language name such as
>  "/de/" by using an additional XML catalog for German, without modificati=

Yes, the DTD path used at this time looks much more consistent, than
just "/share/...". May be it is possible to point onto DTDs in such
way, where HTTP server will return it on request. For example, via
redirect directives, HTML redirects, or symbolic links...

>  of XML files themselves.  I am planning to implement separation of
>  localized part of XSLT stylesheets and entity sets with this
>  mechanism (and eliminate DOCLANG hack).

Good to hear. This would be valuable change (notwithstanding the
DOCLANG is not so dirty hack ;)

As I see the NetBSD documentation is going more and more towards to its
own XML format. I'm have nothing against this movement, so as it will
provide more covenient ways for many things (like <port>). But is this
means that we could more freely implement our own (well documented, of
course) extensions?


Content-Type: application/pgp-signature

Version: GnuPG v1.4.1 (NetBSD)